Description:DESCRIPTION
Technical Field
[001] This disclosure relates generally to Generative Artificial Intelligence (AI), and more particularly to method and system for automatically generating a personalized training curriculum for a user.
Background
[002] In recent years, advancements in Artificial Intelligence (AI), particularly generative AI (also referred to as GenAI), have significantly transformed modern enterprise operations across industries. Generative AI has demonstrated strong capabilities in performing a wide range of tasks, particularly those at foundational or entry level, commonly referred to as “L1” level tasks, such as providing initial customer support, performing entry-level programming, and executing routine data processing. By automating these entry-level tasks, organizations have achieved substantial improvements in operational efficiency, cost savings, and productivity. Consequently, organizations are increasingly adopting generative AI-driven solutions to streamline workflows, reduce turnaround times, and optimize resource utilization.
[003] While the integration of generative AI into organization workflows offers significant benefits, it may also introduce new challenges, particularly with respect to workforce development and long-term skill sustainability. The automation of entry level tasks within a given domain inadvertently limits opportunities for new hires and junior employees to gain the practical, hands-on experience necessary to build a deep understanding of core processes. Without this exposure, employees may struggle to develop the critical thinking and problem-solving skills required for advanced level or more complex tasks (i.e., L2 level tasks), such as advanced programming, system design, or high-level support functions. This growing dependence on generative AI for entry-level tasks poses a significant risk to organizations’ talent pipelines and overall operational resilience. Over time, the diminished opportunities for experiential learning at the entry level can create a critical skills gap, weakening the progression path from L1 to L2 level task competencies. Such a skills gap may result in a shortage of skilled personnel capable of handling complex issues, maintaining legacy systems, or driving innovation at higher operational tiers. Therefore, while generative AI continues to deliver efficiency gains, there is a pressing need for solutions that balance automation benefits with sustained human skill development and knowledge retention within the organization’s ecosystem.
[004] Thus, the existing approaches in the present state of the art fail to address the problem of automatically generating a personalized training curriculum for a user.
SUMMARY
[005] In one embodiment, a method for automatically generating a personalized training curriculum for a user is disclosed. In one example, the method may include receiving information corresponding to a plurality of entry-level tasks executed on a job from a Large Language Model (LLM) at a predefined time interval. The method may include processing the information corresponding to the plurality of entry-level tasks to determine a difficulty score for each of the plurality of entry-level tasks based on a set of parameters. The method may include generating a plurality of semantic clusters comprising one or more of the plurality of entry-level tasks, based on the difficulty score determined for each entry-level task, using a pre-defined clustering technique. The method may include arranging each of the plurality of semantic clusters in a sequential learning order. The method may include generating the personalized training curriculum comprising a set of assignments for the user based on the sequential learning order of each of the plurality of semantic clusters. The method may include rendering the personalized training curriculum via a Graphical User Interface (GUI) to the user.
[006] In another embodiment, a system for automatically generating a personalized training curriculum for a user is disclosed. The system may include a processor and a memory communicatively coupled to the processor. The memory may store processor-executable instructions, which, on execution, may cause the processor to receive information corresponding to a plurality of entry-level tasks executed on a job from a Large Language Model (LLM) at a predefined time interval. The processor-executable instructions, on execution, may further cause the processor to process the information corresponding to the plurality of entry-level tasks to determine a difficulty score for each of the plurality of entry-level tasks based on a set of parameters. The processor-executable instructions, on execution, may further cause the processor to generate a plurality of semantic clusters comprising one or more of the plurality of entry-level tasks, based on the difficulty score determined for each entry-level task, using a pre-defined clustering technique. The processor-executable instructions, on execution, may further cause the processor to arrange each of the plurality of semantic clusters in a sequential learning order. The processor-executable instructions, on execution, may further cause the processor to generate the personalized training curriculum comprising a set of assignments for the user based on the sequential learning order of each of the plurality of semantic clusters. The processor-executable instructions, on execution, may further cause the processor to render the personalized training curriculum via a Graphical User Interface (GUI) to the user.
[007] It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the invention, as claimed.

BRIEF DESCRIPTION OF THE DRAWINGS
[008] The accompanying drawings, which are incorporated in and constitute a part of this disclosure, illustrate exemplary embodiments and, together with the description, serve to explain the disclosed principles.
[009] FIG. 1 is a block diagram of an exemplary system configured for automatically generating a personalized training curriculum for a user, in accordance with some embodiments of the present disclosure.
[010] FIG. 2 illustrates a functional block diagram depicting various modules present within a memory of a computing device configured for automatically generating a personalized training curriculum for a user, in accordance with some embodiments of the present disclosure.
[011] FIG. 3 illustrates a flow diagram of a method for automatically generating a personalized training curriculum for a user, in accordance with some embodiments of the present disclosure.
[012] FIG. 4 illustrates a flow diagram of a method for arranging a plurality of sequential clusters, in accordance with some embodiments of the present disclosure.
[013] FIG. 5 illustrates a flow diagram of a method for monitoring a performance of a user, in accordance with some embodiments of the present disclosure.
[014] FIG. 6 illustrates a flow diagram of a detailed process for determining a difficulty score for each of a plurality of entry-level tasks, in accordance with some embodiments of the present disclosure.
[015] FIG. 7 illustrates a flow diagram of a detailed process for creating a mapping of keywords to at least one domain, in accordance with some embodiments of the present disclosure.
[016] FIG. 8 illustrates a flow diagram depicting a detailed process for arranging a plurality of semantic clusters, in accordance with some embodiments of the present disclosure.
[017] FIG. 9 is a flow diagram depicting a detailed process for monitoring performance of a user, in accordance with some embodiments of the present disclosure.
[018] FIG. 10 is a block diagram of an exemplary computer system for implementing embodiments consistent with the present disclosure.

DETAILED DESCRIPTION
[019] Exemplary embodiments are described with reference to the accompanying drawings. Wherever convenient, the same reference numbers are used throughout the drawings to refer to the same or like parts. While examples and features of disclosed principles are described herein, modifications, adaptations, and other implementations are possible without departing from the spirit and scope of the disclosed embodiments. It is intended that the following detailed description be considered as exemplary only, with the true scope and spirit being indicated by the following claims.
[020] Referring now to FIG. 1, a block diagram of an exemplary system 100 configured for automatically generating a personalized training curriculum for a user is illustrated, in accordance with some embodiments of the present disclosure. As depicted via the present FIG. 1, the system 100 may include a computing device 102. The computing device 102 may be configured to automatically generate the personalized training curriculum for the user. In an embodiment, the personalized training curriculum may correspond to a dynamically generated learning plan tailored to an individual user’s current skill level, role, learning pace, and identified knowledge gaps. The user may refer to an individual, such as an employee, a trainee, or a learner, who interacts with the computing device 102 to access the personalized training curriculum.
[021] In order to automatically generate the personalized training curriculum for the user, initially, the computing device 102 may be configured to receive information corresponding to a plurality of entry-level tasks executed on a job from a Large Language Model (LLM). An entry-level task may correspond to a foundational activity or a basic activity within a specific job domain that requires limited experience or expertise to provide solution. Further, a job may refer to a specific role, a position, or a function performed within an organization that includes tasks (questions) or responsibilities across various complexities. Examples of the plurality of entry-level tasks may include responding to basic customer support queries, performing data entry or validation, writing simple code modules or scripts, generating routine reports, conducting initial software testing, resolving low-complexity technical issues using predefined procedures, and the like. Further, examples of the LLM may include, but are not limited to, a Generative Pre-trained Transformer (GPT) 5, a Claude, a Gemini, a Large Language Model Meta AI (LLaMA), and a Mistral.
[022] In an embodiment, the information corresponding to the plurality of entry-level tasks may include a domain associated with each task, a solution associated with each task, an estimated completion time for each task, a LLM confidence score for the solution associated with each task, and a set of resources required for each task. The computing device 102 may receive the information corresponding to the plurality of entry-level tasks at a predefined time interval, for example, 30 minutes, 1 hour, and the like. In an embodiment, the domain may correspond to a specific area or a field to which an entry-level task may belong, e.g., web development, customer support, data management, and the like. The solution associated with each task may represent a method, steps, or output provided to complete a given entry-level task. For example, when an entry-level task is to write a reply to a customer asking for refund status, a solution may include a predefined message template explaining refund timelines and process. The estimated completion time for each task may indicate an approximate time required by the user to complete the entry-level task. Further, the LLM confidence score may reflect a level of confidence that the LLM has in the correctness of its generated solution. For example, the LLM confidence score for a solution generated by the LLM for an entry-level task, e.g., generate a Structured Query Language (SQL) query to retrieve customer data, may be 85%. Further, the set of resources required for each task may include tools, datasets, documents, Application Programming Interfaces (APIs), or learning materials necessary to perform each task.
[023] Upon receiving the information corresponding to the plurality of entry-level tasks, the computing device 102 may be configured to process the information corresponding to the plurality of entry-level tasks to determine a difficulty score for each of the plurality of entry-level tasks based on a set of parameters. The set of parameters, for example, may include an expected time to complete each task, a number of questions historically asked to complete each task, and the pre-trained LLM confidence score for the solution associated with each task. In an embodiment, the difficulty score may represent a complexity of each task. Once the difficulty score is determined, the computing device 102 may be configured to generate a plurality of semantic clusters including one or more of the plurality of entry-level tasks, based on the difficulty score determined for each entry-level task. The plurality of semantic clusters may be generated using a pre-defined clustering technique. A semantic cluster may include entry-level tasks that are grouped together based on a corresponding difficulty score. For example, each entry-level task having a difficulty score ranging between 85% to 100% may be grouped together in one semantic cluster. Further, the pre-defined clustering technique, for example, may include K-Means clustering, hierarchical clustering, Density-Based Spatial Clustering of Applications with Noise (DBSCAN), semantic embedding-based clustering, and the like.
[024] Upon generating the plurality of semantic clusters, the computing device 102 may be configured to arrange each of the plurality of semantic clusters in a sequential learning order. In an embodiment, the sequential learning order may correspond to an arrangement of the plurality of semantic clusters into a deterministic linear sequence specifying a progression from prerequisite or lower‑complexity clusters to subsequent clusters of greater complexity, such that each later cluster is built upon knowledge or context established by preceding clusters to facilitate cumulative understanding or skill acquisition. Further, the computing device 102 may be configured to generate the personalized training curriculum for the user based on the sequential learning order of each of the plurality of semantic clusters. In an embodiment, the personalized training curriculum may include a set of assignments. An assignment may correspond to a user-specific learning module that is generated for a specific user and is linked to one semantic cluster in the sequential learning order. Further, each assignment may include a set of tasks (also referred to as a set of questions) to be performed by the user. The set of tasks for the assignment may be selected to assess, reinforce, or enhance the user’s understanding of one or more concepts, domains, or skills associated with a corresponding semantic cluster.
[025] For example, in a software development domain, an assignment may correspond to a ‘Version Control Basics’ module generated from a semantic cluster associated with version management concepts. The set of tasks within the assignment, i.e., ‘Version Control Basics’ module may include, initializing a Git repository, committing changes to local branches, merging branches and resolving conflicts, and pushing updates to a remote repository. By way of another example, in a data analytics domain, an assignment may correspond to a ‘Data Cleaning and Preprocessing’ module, where the set of tasks may include, identifying and handling missing values in a dataset, normalizing numerical attributes, encoding categorical variables, and generating summary statistics.
[026] Once the personalized training curriculum is generated, the computing device 102 may be configured to render the personalized training curriculum to the user via a Graphical User Interface (GUI), e.g., a user interface 110 of the computing device 102 or a user interface of an external device of one or more external devices 118. In particular, rendering the personalized training curriculum to the user may include rendering an assignment from the set of assignments within the personalized training curriculum to the user. This complete method of automatically generating the personalized training curriculum for the user is further explained in detail in conjunction with FIG. 2 and FIG. 9.
[027] Examples of the computing device 102 may include, but is not limited to, a mobile phone, a laptop, a desktop, or a Personal Digital Assistant (PDA), an application server, and so forth. The computing device 102 may further include a memory 104, a processor 106, and an Input/Output unit 108. The I/O unit 108 may further include the user interface 110. The user (e.g., the trainee, the new hire, etc.) or an administrator (e.g., a system administrator) may interact with the computing device 102 and vice versa through the I/O unit 108.
[028] The I/O unit 108 may be used to display results (i.e., the plurality of entry-level tasks, the assignment from the set of assignments, the personalized training curriculum, etc.) based on actions performed by the computing device 102, to the user. The user interface 110 may be used by the user to provide inputs to the computing device 102. Thus, for example, in some embodiments, the computing device 102 may ingest the user input that includes a user response corresponding to the set of tasks provided in the assignment. Further, for example, in some embodiments, the computing device 102 may render intermediate results (e.g., the plurality of entry-level tasks, the difficulty score for each of the plurality of entry-level tasks, etc.) or final results (e.g., the personalized training curriculum) to the user via the user interface 110.
[029] The memory 104 may store instructions that, when executed by the processor 106, may cause the processor 106 to automatically generate the personalized training curriculum for the user. As will be described in greater detail in conjunction with FIG. 2 to FIG. 9, in order to automatically generate the personalized training curriculum, the processor 106 in conjunction with the memory 104 may perform various functions including receiving information corresponding to the plurality of entry-level tasks, processing the information corresponding to the plurality of entry-level tasks to determine the difficulty score, generating the plurality of semantic clusters, arranging each of the plurality of semantic clusters in the sequential learning order, generating the personalized training curriculum, rendering the personalized training curriculum, and the like.
[030] The memory 104 may also store various data (e.g., the plurality of entry-level tasks, the difficulty score for each of the plurality of entry-level tasks, the plurality of semantic clusters, the sequential learning order of each of the plurality of semantic clusters, and the like) that may be captured, processed, and/or required by the computing device 102. The memory 104 may be a non-volatile memory (e.g., flash memory, Read Only Memory (ROM), Programmable ROM (PROM), Erasable PROM (EPROM), Electrically EPROM (EEPROM) memory, etc.) or a volatile memory (e.g., Dynamic Random Access Memory (DRAM), Static Random-Access memory (SRAM), etc.).
[031] Further, the computing device 102 may interact with a server 112 or the one or more external devices 118 over a network 116 for sending and receiving various data. The network 116, for example, may be any wired or wireless communication network and the examples may include, but may be not limited to, the Internet, Wireless Local Area Network (WLAN), Wi-Fi, Long Term Evolution (LTE), Worldwide Interoperability for Microwave Access (WiMAX), and General Packet Radio Service (GPRS).
[032] In an embodiment, the computing device 102 may fetch the plurality of entry-level tasks executed on the job from the LLM residing on the server 112. In addition, the server 112 may provide information, such as details about the plurality of entry-level tasks, etc., to the computing device 102 at the pre-defined time interval. The server 112 may further include a database 114. By way of an example, the database 114 may store information corresponding to a plurality of entry-level tasks executed on a plurality of jobs. The database 114 may be periodically updated based on new entry-level task or a new job. Alternatively, the computing device 102 may receive input from the user via the one or more external devices 118. Examples of the one or more external devices 118 may include a laptop, a desktop, a smartphone, a tablet, and the like. This complete process followed by the system 100 is further explained in detail in conjunction with FIG. 2 to FIG. 9.
[033] Referring now to FIG. 2, a functional block diagram 200 depicting various modules present within the memory 104 of the computing device 102 configured for automatically generating the personalized training curriculum for the user is illustrated, in accordance with some embodiments of the present disclosure. FIG. 2 is explained in conjunction with FIG. 1. As depicted via FG. 2, the block diagram 200 may include two modules such as a knowledge extractor and ranker module 204 and a tutor module 206. In an embodiment, each module may be present within the memory 104 of the computing device 102. Further, the block diagram 200 may include a trained LLM agent 202 and a study environment module 208.
[034] In order to automatically generate the personalized training curriculum for the user, initially, the trained LLM agent 202 may be configured to provide the information corresponding to the plurality of entry-level tasks executed on the job to the knowledge extractor and ranker module 204. In an embodiment, the trained LLM agent 202 may serve as a foundational knowledge base configured with domain-specific skills and expertise necessary to successfully execute the plurality of entry-level tasks (also referred to as L1 level tasks) associated with one or more jobs. The trained LLM agent 202 may be configured to distill and encapsulate procedural knowledge, task execution strategies, and contextual understanding derived from the plurality of entry-level tasks. The distilled knowledge (i.e., the information) provided by the trained LLM agent 202 may serve as reference material and a baseline dataset for the generation of the personalized training curriculum for the user. In an embodiment, the trained LLM agent 202 may be a specialized LLM (i.e., the LLM), meticulously trained to perform a wide range of the plurality of entry-level tasks successfully. Examples of the LLM may include, but are not limited to, GPT 5, Claude, Gemini, LLaMA, and Mistral. In an embodiment, the trained LLM agent 202 may send the information corresponding to the plurality of entry-level tasks to the knowledge extractor and ranker module 204 after an expiry of the predefined time interval, for example, 30 minutes, 1 hour, and the like. In an embodiment, the information corresponding to the plurality of entry-level tasks may include the domain associated with each task, the solution associated with each task, the estimated completion time for each task, the LLM confidence score for the solution associated with each task, and the set of resources required for each task.
[035] Upon receiving the information corresponding to the plurality of entry-level tasks, the knowledge extractor and ranker module 204 may be configured to process the information corresponding to the plurality of entry-level tasks to determine the difficulty score for each of the plurality of entry-level tasks. The knowledge extractor and ranker module 204 may process the information based on the set of parameters. The set of parameters, for example, may include the expected time to complete each task, the number of questions historically asked to complete each task, and the LLM confidence score for the solution associated with each task. In particular, the knowledge extractor and ranker module 204 may serve as a core distillation and organization module within the computing device 102. The knowledge extractor and ranker module 204 may operating on the knowledge base (i.e., the information) provided by the trained LLM agent 202 to automatically generate a comprehensive and ranked training curriculum, i.e., the personalized training curicullam for the user. In an embodiment, the knowledge extractor and ranker module 204 may perform one or more functions, such as identifiying entry-level tasks and content, performing difficulty ranking and assessment, performing semantic clustering of the entry-level tasks, and identifying important keywords and topics for each semantic cluster.
[036] In particular, identifiying entry-level tasks and content may include identification of the plurality of entry-level tasks based on the information received from the trained LLM agent 202. In other words, the knowledge extractor and ranker module 204 may be configured to identify an exhaustive set of L1-level tasks (i.e., the plurality of entry-level tasks), along with content that includes corresponding critical skills, required resources, and detailed step-by-step solutions based on the information received from the trained LLM agent 202. In particular, the knowledge extractor and ranker module 204 may identify the domain associated with each task, the solution associated with each task, the estimated completion time for each task, the LLM confidence score for the solution associated with each task, and the set of resources required for each task.
[037] Further, to perform difficulty ranking and assessment, the knowledge extractor and ranker module 204 may be configured to determine the difficulty score for each of the plurality of entry-level tasks based on the set of parameters. The set of parameters may include the expected time to complete each task, the number of questions historically asked to complete each task, and the LLM confidence score for the solution associated with each task. In other words, the knowledge extractor and ranker module 204 may be configured to assign the difficulty score to each entry-level task based on a multi-factor analysis. In one embodiment, the difficulty score may be determined by computing a weighted sum of multiple parameters (i.e., the set of parameters). For example, the multiple parameters may include the expected time to complete each task that is determined by the trained LLM agent 202 or an external software executing the trained LLM agent 202, a degree of clarification required by the the trained LLM agent 202, as indicated by a number or frequency of follow-up questions during task execution (i.e., the number of questions historically asked), and a confidence score (i.e., the LLM confidence score for the solution) of the trained LLM agent 202, which may be derived from token probability metrics (for example, low log-probability or high perplexity values). In some embodiments, the difficulty score corresponding to each entry-level task may be provided by the trained LLM agent 202.
[038] Further, the knowledge extractor and ranker module 204 may be configured to generate the plurality of semantic clusters including the one or more of the plurality of entry-level tasks, based on the difficulty score determined for each entry-level task. The plurality of semantic clusters may be generated using the pre-defined clustering technique. In particular, the knowledge extractor and ranker module 204 may be configured to group the plurality of entry-level tasks into the plurality of semantic clusters based on their contextual and linguistic similarity. In one embodiment, the clustering of the plurality of entry-level tasks may be performed using the K-means clustering (i.e., the pre-defined clustering technique) applied to vector embeddings of each entry-level task description. As a result, semantically related entry-level tasks may be grouped within the same semantic cluster. Once the plurality of semantic clusters is generated, the knowledge extractor and ranker module 204 may be configured to arrange the plurality of semantic clusters in the sequential learning order. To arrange the plurality of semantic clusters in the sequential learning order, the knowledge extractor and ranker module 204 may identify a first cluster with a minimum number of domains, from the plurality of semantic clusters. The first cluster may be the simplest cluster used to establish the foundational learning base covering fewest domain.
[039] Further, the knowledge extractor and ranker module 204 may iteratively select a subsequent cluster with a minimum number of new domains not previously introduced in the first cluster. In other words, the knowledge extractor and ranker module 204 may repeatedly choose the next cluster that introduces fewest new domains not already covered in any previous cluster. This may ensure that each new cluster builds smoothly on previously learned topics, creating a logical learning progression. The knowledge extractor and ranker module 204 may repeat the selection of the subsequent cluster until each of the plurality of semantic clusters are arranged in the sequential learning order. In an embodiment, a number of domains for each cluster of the plurality of semantic clusters is determined by identifying one or more keywords from each cluster and mapping each of the one or more keywords to at least one domain. Further, the knowledge extractor and ranker module 204 may assign a training time to each cluster based on the number of domains introduced by a corresponding cluster. By way of an example, suppose a cluster A has keywords mapping to two domains (e.g., ‘payments’ and ‘security’), and a cluster B has keywords mapping to three domains (e.g., ‘filter’, ‘query’, and ‘database basic’). In this case, the cluster A may be assigned a training time of 30 minutes, and the cluster B may be assigned a training time of 45 minutes.
[040] In particular, the knowledge extractor and ranker module 204 may be configured to identify significant keywords and corresponding topics (i.e., domains) for each of the plurality of semantic clusters. To identify significant keywords and corresponding topics, the knowledge extractor and ranker module 204 may identify common themes or frequently recurring terms, subtasks, or descriptions that appears in multiple entry-level tasks. Further, the knowledge extractor and ranker module 204 may map the identified keywords to one or more topics using a keyword-to-topic mapper, implemented using a predefined taxonomy or a machine learning (ML) model. By using the keyword-to-topic mapper, the knowledge extractor and ranker module 204 may identify one or more important topics for each semantic cluster based on keywords associated with that semantic cluster. A taxonomy is a structured framework or a classification system used to organize information, concepts, or entities into hierarchical categories based on their relationships or characteristics. By way of an example, in a software engineering taxonomy, tasks may be categorized under hierarchical levels such as programming, web development, front-end development, etc. In an embodiment, each topic may represent a coherent knowledge entity that is sufficiently independent to support targeted training. In particular, a topic (e.g., a domain) may refer to a distinct and self-contained unit of knowledge that is sufficiently independent in its scope and content. In some embodiments, each topic may have an associated training manual or knowledge resource maintained internally within an organization. For more general or widely recognized topics, publicly available resources or standardized taxonomies may be utilized.
[041] Once the plurality of semantic clusters are arranged, the tutor module 206 may be configured to generate the personalized training curriculum for the user. The personalized training curriculum for the user may be generated based on the sequential learning order of each of the plurality of semantic clusters. In particular, the tutor module 206 may function as a primary orchestration and personalization module, configured to deliver an adaptive and highly optimized training curriculum to the user. To generate the adaptive and highly optimized training curriculum, the tutor module 206 may utilize a structured knowledge base (i.e., the plurality of semantic clusters) generated by the knowledge extractor and ranker agent. In an embodiment, the tutor module 206 may be capable of delivering adaptive and highly optimized training curriculum by generating a personalized sequence of assignments by selecting questions (i.e., tasks) with difficulty levels adjusted in real time based on one or more performance parameters. The one or more performance parameters may include total marks obtained by the user for an assignment, a total time taken by the user to complete the assignment, and a completion status of the assignment. In some embodiments, the difficulty level of the subsequent assignment rendered to the user may be adjusted based on the user's performance metrics, the complexity of previously completed assignments, and the available training duration.
[042] Further, the tutor module 206 may be capable of provisioning one or more contextual learning resources, such as relevant company documents, external knowledge bases, other supplementary materials, etc., to the user. In particular, the tutor module 206 may be capable of analyzing the performance of the user and the nature of a current task to determine when to supply the one or more contextual learning resources to the user, thereby ensuring that the user can successfully complete the assignment. In addition, the tutor module 206 may be capable of monitoring performance of the user and provide feedback. The tutor module 206 may be configured to continuously monitors the user's performance through interactions of the user with the computing device 102 via the study environment module 208. The tutor module 206 may use this performance data to evaluate task completion, track progress, and provide constructive feedback, to refine the the difficulty level and composition of subsequent assignments. In particular, the tutor agent 206 may deliver a training experience that is dynamic, individualized, and responsive to evolving needs and capabilities of the user, rather than a static or uniform instructional approach.
[043] In particular, for the purpose of generating the personalized training curriculum for the user, the tutor module 206 may employ a two-stage process that begins with determining an ordered sequence (i.e., the sequenctial learning ordering) of the plurality of semantic clusters followed by selecting an appropriate set of questions from each seamntic cluster. The plurality of semantic clusters may be generated by the knowledge extractor and ranker module 204. Further, the determination of a cluster sequence (i.e., the sequenctial learning order) may be performed using a pre-defined procedure, which is designed to regulate the progression of difficulty of the assignments throughout a training schedule. For example, the pre-defined procedure may include the sequencing of the plurality of semantic clusters by selecting the first cluster associated with the minimum number of topics (i.e., the minimum number of domains). Further, a subsequent cluster is then selected based on a criterion that the subsequent cluster may introduce a minimum number of new topics not previously covered by any earlier-selected cluster. This selection process is repeated until all the plurality of semantic clusters have been incorporated into the sequence, with any ties resolved arbitrarily.
[044] Once the sequenctial learning order for the plurality of semantic clusters is established, the training time is assigned to each cluster in proportion to the number of new topics introduced by that corresponding cluster. In an embodiment, based on the training time allocated to each cluster, the tutor module 206 may select a greater number of questions or questions having higher difficulty ratings for clusters with longer assigned durations, ensuring that question complexity aligns with the training time available for learning within each cluster. This approach ensures a progressively structured learning path with controlled introduction of new subject matter. Further, this predefined procedure may leverage a novel keyword and topic-based framework to ensure that learning complexity increases in a controlled manner and that required learning resources are introduced precisely when needed. This approach further guarantees that the questions (tasks) selected from each cluster exhibit conceptual continuity, allowing subsequent assignments to build logically upon the knowledge gained from earlier assignments.
[045] Using the above-described selection process for each cluster, the tutor module 206 may generate the personalized training curriculum for the user. In an embodiment, the tutor module 206 may further adapt the personalized training curriculum in real time based on the performance of the user by employing another predefined procedure, as described herein. According to this predfined procedure, the tutor module 206 may initialize various parameters (i.e., the one or more performance parameters) including the total marks obtained by the user, the total time taken by the user, the completion status, the remaining expected training time, an initial difficulty level, and a performance log for recording task-level outcomes. The tutor module 206 may then iteratively processes each cluster based on the sequential learning order, where each cluster is handled within its allocated training time. For each cluster, the tutor module 206 may initialize a timer and provide the one or more contextual learning resources corresponding to new topics (i.e., new domains) introduced by that cluster.
[046] Further, the tutor module 206 may select assignments from a current cluster at the difficulty level corresponding to the most recent performance of the user and presents each task to the user via the study environment module 208. Further, the tutor module 206 may receive the total marks obtained and the time taken to complete the task from the study environment module 208. Based on the total marks and the total time, the tutor module 206 may adjust the difficulty level for the subsequent assignment by raising the difficulty level of the subsequent assignment when performance exceeds a minimum predefined threshold (e.g., 40%) or lowering the difficulty level when the performance falls below the minimum predefined threshold. Further, the tutor module 206 may append this performance result of the user to a performance log. The tutor module 206 may then continue selecting assignments until the allocated time for the current cluster is exhausted, after which the tutor module 206 may proceed to the next cluster. This adaptive process continues until all clusters have been covered, thereby generating a personalized and responsive training experience tailored to evolving learning needs of the user.
[047] The study environment module 208 may function as a primary user interface (e.g., the user interface 110) of the computing device 102, enabling direct interaction between the user and the computing device 102 (multi-agent training framework). In some embodiments, the study environment module 208 may be the user interface of the one or more external device 118. As a user-facing component, the study environment module 208 may be configured to provide a practical and intuitive learning environment. In one embodiment, the study environment module 208 may be configured to present the user with a complete description of each hands-on assignment or task, including the necessary instructions and contextual information (i.e., the one or more contextual learning resources) required to understand the objectives and execution requirements. The study environment module 208 may further serves as a secure mechanism for receiving and capturing the solutions (i.e., the user response) submitted by the user. The user response may be received, code snippets, document uploads, or text-based answers. Further, the study environment module 208 may transmit the solutions submitted by the user to the tutor agent 206 for evaluation and performance analysis. Additionally, the study environment module 208 may simulate a realistic work environment enabling the user to gain practical, hands-on experience without risking operational disruptions. This direct application of knowledge is critical for transitioning theoretical understanding into actionable skills. In this manner, the study environment module 208 may ensure that the training experience extends beyond theoretical instruction and supports the acquisition of applied competencies aligned with real-world entry-level activities.
[048] It should be noted that all such aforementioned modules 202 – 208 may be represented as a single module or a combination of different modules. Further, as will be appreciated by those skilled in the art, each of the modules 202 – 208 may reside, in whole or in parts, on one device or multiple devices in communication with each other. In some embodiments, each of the modules 202 – 208 may be implemented as dedicated hardware circuit comprising custom application-specific integrated circuit (ASIC) or gate arrays, off-the-shelf semiconductors such as logic chips, transistors, or other discrete components. Each of the modules 202 – 208 may also be implemented in a programmable hardware device such as a field programmable gate array (FPGA), programmable array logic, programmable logic device, and so forth. Alternatively, each of the modules 202 – 208 may be implemented in software for execution by various types of processors (e.g., the processor 106). An identified module of executable code may, for instance, include one or more physical or logical blocks of computer instructions, which may, for instance, be organized as an object, procedure, function, or other construct. Nevertheless, the executables of an identified module or component need not be physically located together but may include disparate instructions stored in different locations which, when joined logically together, include the module and achieve the stated purpose of the module. Indeed, a module of executable code could be a single instruction, or many instructions, and may even be distributed over several different code segments, among different applications, and across several memory devices.
[049] As will be appreciated by one skilled in the art, a variety of processes may be employed for automatically generating the personalized training curriculum for the user. For example, the exemplary system 100 and the associated computing device 102 may automatically generate the test script by the processes discussed herein. In particular, as will be appreciated by those of ordinary skill in the art, control logic and/or automated routines for performing the techniques and steps described herein may be implemented by the system 100 and the associated computing device 102, either by hardware, software, or combinations of hardware and software. For example, suitable code may be accessed and executed by the one or more processors on the system 100 to perform some or all of the techniques described herein. Similarly, application specific integrated circuits (ASICs) configured to perform some, or all of the processes described herein may be included in the one or more processors on the system 100.
[050] Referring now to FIG. 3, a flow diagram of a method 300 for automatically generating the personalized training curriculum for the user is illustrated, in accordance with some embodiments of the present disclosure. FIG. 3 is explained in conjunction with FIGS. 1 and 2. Further, each step of the method 300 may be implemented by various modules present within the memory 104 of the computing device 102.
[051] In order to generate the personalized training curriculum for the user, initially, at step 302, information corresponding to the plurality of entry-level tasks executed on the job may be received from the LLM. The information corresponding to the plurality of entry-level tasks may be received at the predefined time interval, for example, 30 minutes, 1 hour, and the like. In an embodiment, an entry-level task may correspond to a foundational activity or a basic activity within a specific job domain that requires limited experience or expertise to provide a solution for that task. Further, a job may refer to a specific role, a position, or a function performed within an organization that includes tasks or responsibilities across various complexities. Examples of the plurality of entry-level tasks may include responding to basic customer support queries, performing data entry or validation, writing simple code modules or scripts, generating routine reports, conducting initial software testing, resolving low-complexity technical issues using predefined procedures, and the like. Further, examples of the LLM may include, but are not limited to, GPT 5, Claude, Gemini, LLaMA, and Mistral.
[052] In an embodiment, the information corresponding to the plurality of entry-level tasks may include the domain associated with each task, the solution associated with each task, the estimated completion time for each task, the LLM confidence score for the solution associated with each task, and the set of resources required for each task. The domain may correspond to a specific area or a field to which an entry-level task may belong, e.g., web development, customer support, data management, and the like. The solution associated with each task may represent a method, steps, or output provided to complete a given entry-level task. The estimated completion time for each task may indicate an approximate time required by the user to complete the entry-level task. Further, the LLM confidence score may reflect a level of confidence that the LLM has in the correctness of its generated solution. Furthermore, the set of resources required for each task may include tools, datasets, documents, APIs, or learning materials necessary to perform each task.
[053] Upon receiving the information corresponding to the plurality of entry-level tasks, at step 304, the information corresponding to the plurality of entry-level tasks may be processed to determine the difficulty score for each of the plurality of entry-level tasks. In an embodiment, the difficulty score may represent the complexity of each task. The difficulty score corresponding to the plurality of entry-level tasks may be determined based on the set of parameters. The set of parameters, for example, may include the expected time to complete each task, the number of questions historically asked to complete each task, and the LLM confidence score for the solution associated with each task.
[054] Once the difficulty score is determined, at step 306, the plurality of semantic clusters including one or more of the plurality of entry-level tasks may be generated, based on the difficulty score determined for each entry-level task. The plurality of semantic clusters may be generated using the pre-defined clustering technique. A semantic cluster may include entry-level tasks that are grouped together based on the corresponding difficulty score. The pre-defined clustering technique, for example, may include K-Means clustering, hierarchical clustering, DBSCAN, semantic embedding-based clustering, and the like.
[055] Upon generating the plurality of semantic clusters, at step 308, each of the plurality of semantic clusters may be arranged in the sequential learning order. In an embodiment, the sequential learning order may correspond to an arrangement of the plurality of semantic clusters into a deterministic linear sequence specifying a progression from prerequisite or lower‑complexity clusters to subsequent clusters of greater complexity, such that each later cluster is built upon knowledge or context established by preceding clusters to facilitate cumulative understanding or skill acquisition. A method for arranging the plurality of semantic clusters in the sequential learning order is further explained in detail in conjunction with FIG. 4.
[056] Further, at step 310, the personalized training curriculum may be generated for the user. The personalized training curriculum may be generated based on the sequential learning order of each of the plurality of semantic clusters. In an embodiment, the personalized training curriculum may include the set of assignments. An assignment may correspond to a user-specific learning module that is generated for a specific user and is linked to one semantic cluster in the sequential learning order. Further, each assignment may include the set of tasks to be performed by the user. The set of tasks for the assignment may be selected to assess, reinforce, or enhance the user’s understanding of one or more concepts, domains, or skills associated with the corresponding semantic cluster. Once the personalized training curriculum is generated, at step 312, the personalized training curriculum may be rendered to the user via the GUI. With reference to FIG. 1, the GUI may correspond to the user interface 110 of the computing device 102 or the user interface of the external device of the one or more external devices 118. With reference to FIG. 2, the GUI may correspond to the study environment module 208. A method of rendering the personalized training curriculum to the user is further explained in detail in conjunction with FIG. 5.
[057] By way of an example, consider an exemplary scenario where a synthetic dataset is utilized to demonstrate generation of the personalized training curriculum. For illustration, five representative technical topics (domains) are considered, e.g., Operating Systems (OS), Computer Networks (CN), Databases and Structured Query Language (SQL) (DB), Machine Learning (ML), and Artificial Intelligence (AI). Each topic may be associated with a set of domain-specific keywords. For example, an OS domain may include keywords, such as, process, scheduling, memory, concurrency, etc., a CN domain may include keyword, such as packet, latency, routing, etc., a DB domain may include keywords, such as query, table, join, function, memory, transaction, indexing, normalization, etc., a ML domain may include keywords such as model, dataset, training, function, agent, state, optimization, regularization, etc., and an AI domain may include keywords such as agent, state, search, function, resource, planning, reasoning, knowledge, etc.
[058] In an embodiment, to initiate a difficulty level assessment process, each task is assigned the difficulty score based on a weighted combination of three normalized parameters (i.e., the set of parameters), i.e., an expected time (t) to complete the task, the number of questions (e.g., a degree of clarification (c)) historically asked to complete each task, and a model uncertainty (m) derived from the LLM’s confidence score. For example, the difficulty score (D) for the task may be computed as D = w(t)·t + w(c)·c + w(m)·m, where weights (wt=0.4, wc=0.3, wm=0.3) reflect the relative contribution of each parameter. For example, a difficulty score for the task may be determined to be 0.5 based on representative values for t, c, and m.
[059] Further, a semantic clustering operation may be performed in which the plurality of entry-level tasks are grouped into multi-topic clusters (i.e., the pluraity of semantic clusters) according to keyword similarity, resulting in clusters such as a cluster A (memory, resource, concurrency, filesystem) mapped to OS, CN, and DB domains, a cluster B (function, agent, state, regularization) mapped to DB, ML, and AI domains, a cluster C (security, planning, knowledge) mapped to OS, CN, and AI domains, a cluster D (optimization, performance, reasoning) mapped to ML and AI domains, and a cluster E (bandwidth, latency, packet) mapped to the CN domain. Once the plurality of semantic clusters are generated, the plurality of semantic clusters are arramged in the sequential learning order that minimizes the introduction of new topics at each subsequent cluster, yielding a sequence, i.e., the cluster E, the cluster D, the cluster B, the cluster A, and the cluster C with time allocations (i.e., the training time) proportional to the number of newly introduced topics.
[060] Further, each assignment from each cluster is iteratively presented to the user based on the allocated training time and the real-time performance of the user. For each presented assignment, the one or more performance parameters are recorded and the difficulty level for the subsequent assignment may be determined. Once all the set of assignments of a current cluster is executed or the allocated training time is exceeded, the set of assignments corresponding to the next cluster may be rendered to the user. Upon completion of all the plurality of semantic clusters, a final cumulative score may be calculated. Further, the final cumulative score may be updated in the performance log and the difficulty level representing a proficiency of the user may be determined. This example demonstrates that semantic clustering, topic-based sequencing, adaptive assignment selection, and real-time performance monitoring may generate the personalized training curriculum for the user.
[061] Referring now to FIG. 4, a flow diagram of a method 400 for arranging the plurality of sequential clusters is illustrated, in accordance with some embodiments of the present disclosure. FIG. 4 is explained in conjunction with FIGS. 1, 2, and 3. Further, each step of the method 400 may be implemented by various modules present within the memory 104 of the computing device 102 of the system 100.
[062] In order to arrange the plurality of semantic clusters as mentioned via step 308, at step 402, the first cluster with the minimum number of domains (also referred to as topics) may be selected from the plurality of semantic clusters. In other words, a cluster from the plurality of semantic clusters that includes few different domains (i.e., the topics) may be selected first. This is because the cluster covering few different domains may be simpler for a beginner user to understand and initiate learning. For example, consider the plurality of semantic clusters includes three clusters, i.e., a cluster A, a cluster B, and a cluster C. In the three clusters, the cluster A includes two domains (e.g., a domain Y and a domain Z), the cluster B includes one domain (e.g., a domain X), and the cluster C includes three domains (e.g., a domain T, a domain U, and a domain V). In this case, the cluster B having the minimum number of domains, i.e., one domain (domain X) may be selected as the first cluster in the sequential learning order, as it introduces least number of domains.
[063] Upon selecting the first cluster, at step 404, a subsequent cluster with a minimum number of new domains not previously introduced in the first cluster may be iteratively selected. In continuation to the above example, once the cluster B including the domain X is selected as the first cluster, the remaining clusters (i.e., the cluster A and the cluster C) may be analyzed to identify the subsequent cluster that introduces the minimum number of new domains not previously introduced in the cluster B. In this example, based on the analysis of the remaining clusters, since the cluster A introduces two new domains while the cluster C introduces three new domains, the cluster A may be selected as the subsequent cluster having the minimum number of new domains not previously introduced in the cluster B.
[064] Further, the selection of each of the plurality of semantic clusters may be repeated until each of the plurality of semantic clusters are arranged in the sequential learning order, as mentioned via at step 406. In other words, the selection of the subsequent cluster may be iteratively performed until all of the plurality of semantic clusters have been arranged in the sequential learning order. In conituation to above example, the selection of the subsequent cluster is iteratively done until all the three clusters, i.e., the cluster A, the cluster B, and the cluster C are arranged in the sequential learning order. In an embodiment, the number of domains for each cluster of the plurality of semantic clusters may be determined by identifying the one or more keywords from each cluster and mapping each of the one or more keywords to at least one domain.
[065] In continuation to above example, suppose the domain X for the cluster B may be identified based on keywords such as ‘entry,’ ‘typing,’ and ‘forms’ mapping to domain X. Similarly, the domain Y and the domain Z may be identified for the cluster A based on mapping of keywords such as ‘spreadsheet,’ ‘formula,’ and ‘cells’ to the domain Y and the domain Z. Additionally, the domain T, the domain U, and the domain V may be identified based on keywords such as ‘chart,’ ‘dashboard,’ and ‘reporting’ mapping to the domain T, the domain U, and the domain V. Accordingly, the cluster B, the cluster A, and the cluster C may be determined to include one domain, two domains, and three domains.
[066] Once the plurality of semantic clusters and the number of domains corresponding to the plurality of semantic clusters are identified, the training time may be allocated to each cluster. The training time may be allocated based on the number of domains introduced by the corresponding cluster. In an embodiment, the training time may refer to a duration or a time interval assigned to a semantic cluster for user learning and task completion. The allocated training time may correspond to the complexity of the semantic cluster, including the number of domains it introduces. By way of an example, once the cluster B, the cluster A, and the cluster C are determined to include one domain, two domains, and three domains respectively, the training time may be allocated to the corresponding cluster proportional to the number of domains introduced by each cluster. For instance, the cluster B may be allocated a shorter training time (e.g., 1 hour) as it includes one domain. Further, the cluster A may be allocated a moderate training time (e.g., 2 hours) and the cluster C may be allocated a longer training time (e.g., 3 hours), as the cluster A and the cluster C may include two domains and three domains, respectively.
[067] Referring now to FIG. 5, a flow diagram of a method 500 for monitoring the performance of the user is illustrated, in accordance with some embodiments of the present disclosure. FIG. 5 is explained in conjunction with FIGS. 1, 2, 3 and 4. Further, each step of the method 500 may be implemented by various modules present within the memory 104 of the computing device 102 of the system 100.
[068] To monitor the performance of the user corresponding to the personalized training curriculum, at step 502, the assignment from the set of assignments present within the personalized training curriculum may be rendered to the user. In an embodiment, the assignment may be rendered to the user via the GUI (e.g., the user interface 110) of the computing device 102. In some embodiments, the assignment may be rendered to the user via the user interface of the external device of the one or more external devices 118. With reference to FIG. 2, the GUI may correspond to the study environment module 208. In other words, the study environment module 208 may be part of either the computing device 102 or the external device. In an embodiment, the assignment may include the set of tasks. The set of tasks for the assignment may be selected to assess, reinforce, or enhance the user’s understanding of one or more concepts, domains, or skills associated with a corresponding semantic cluster.
[069] By way of an example, if an assignment corresponds to the cluster B, which includes a set of tasks associated with the domain X (e.g., data entry domain), the set of tasks in the assignment may include activities such as entering customer information from a sample form into a digital template, correcting formatting errors in a partially completed data-entry sheet, and validating sample entries for accuracy by comparing them with a provided reference sheet.
[070] In response to rendering the assignment to the user, at step 504, the user response may be received corresponding to each of the set of tasks provided in the assignment. Upon receiving the user response, at step 506, the performance of the user may be monitored. In an embodiment, the performance of the user may be monitored based on the received user response and the one or more performance parameters. The one or more performance parameters, for example, may include the total marks obtained by the user for the assignment, the total time taken by the user to complete the assignment, and the completion status of the assignment by the user.
[071] For example, after the user completes the set of tasks in the assignment corresponding to the cluster B, the user response for each task along with the one or more performance parameters may be used to monitor the user’s performance. For instance, if the user provides correct responses for each task and completes the set of tasks within the expected time, a high-performance score (e.g., 99%) indicating timely and accurate completion may be assigned to the user. Conversely, if the user makes several entry errors or exceeds the expected time, a lower performance score (e.g., 45%) may be assigned to the user.
[072] Further, based on the performance of the user, at step 508, the difficulty level of the subsequent assignment to be rendered to the user may be adjusted. For instance, if the user achieves the high-performance score (e.g., 99%), the difficulty level of the next assignment (i.e., the subsequent assignment) may be increased by selecting more complex tasks from the next cluster in the sequential learning order. Conversely, if the user receives the low performance score (e.g., 45%), the difficulty level of the set of tasks for the subsequent assignment may be reduced by selecting simpler tasks, providing additional guidance, or assigning remedial tasks before progressing to the next cluster. This may ensure that the subsequent assignment is appropriately aligned with the skill level of the user.
[073] In an embodiment, the one or more contextual learning resources may be identified based on the performance of the user and the nature of the current task. The one or more contextual learning resources, for example, may include relevant company documents, external knowledge bases, other supplementary materials, etc. Further, the one or more contextual learning resources may be displayed to the user via the GUI at a predefined trigger during an execution of the current task. For example, the predefined trigger may include 5 failed attempts to respond to a task of the set of tasks. In other words, after 5 wrong user responses for the task, the one or more contextual learning resources may be displayed to the user. In an embodiment, the predefined trigger may be defined by the system administrator or an employer.
[074] In continuation to the above example, when the user is working on a task, i.e., entering customer information from the sample form into the digital template, of the cluster B. In this case, during execution of the task, if the user is repeatedly making entry errors or taking longer than the expected time, the one or more contextual learning resources may be identified for the user based on the performance of the user and the nature of the current task. The one or more contextual learning resources may include, for instance, a company’s data-entry guideline document, an external tutorial on accurate form transcription, or other supplementary reference materials. Further, after the predefined trigger (e.g., 5 failed attempts), the one or more contextual learning resources may be rendered to the user, thereby assisting the user in completing the task more effectively.
[075] Referring now to FIG. 6, a flow diagram of a detailed process 600 for determining the difficulty score for each of the plurality of entry-level tasks is illustrated, in accordance with some embodiments of the present disclosure. FIG. 6 is explained in conjunction with FIGS. 1, 2, 3, 4 and 5. Further, each step of the method 600 may be implemented by various modules present within the memory 104 of the computing device 102.
[076] At step 602, the process 600 for determining the difficulty score for each of the plurality of entry-level tasks may start. To determine the difficulty score, initially, at step 604, a task (i.e., an entry-level task of the plurality of entry-level task) may be received as an input. In other words, the information corresponding to the task may be received as the input. The information may include the domain associated with the task, the solution associated with the task, the estimated completion time for the task, the LLM confidence score for the solution associated with the task, and the set of resources required for the task. Upon receiving the input, at step 606, the set of parameters correponding to the task may be estimated. The set of parameters may include the expected time to complete the task (i.e., a historical average time required to complete the task), a number of questions historically asked to complete the task (i.e., a degree of clarification historically needed to complete the task), and the LLM confidence score for the solution associated with the task (i.e., a model confidence score). As depicted via step 606, the historical average time, the degree of clarification, and the model confidence score are represented via a variable ‘t’, ‘d’, and ‘c’, respectively. Once the set of parameters are estimated, at step 608, the difficulty score for the task may be computed by applying a weighted combination of the set of parameters as depicted via an equation 1.
Difficulty score = w₁t + w₂d + w₃c … (1)

[077] In equation 1, w₁, w₂, and w₃ may represent predetermined weighting coefficients that may be fixed or dynamically adjusted. Once the difficulty score for the task is computed, at step 610, the difficulty score computed for the task is stored for further processing, e.g., for generating the plurality of semantic clusters. Further, the process 600 may terminate at step 612.
[078] Referring now to FIG. 7, a flow diagram of a detailed process 700 for creating a mapping of keywords to at least one domain is illustrated, in accordance with some embodiments of the present disclosure. FIG. 7 is explained in conjunction with FIGS. 1, 2, 3, 4, 5 and 6. Further, each step of the method 700 may be implemented by various modules present within the memory 104 of the computing device 102 of the system 100.
[079] At step 702, the process 700 for creating the mapping of the keywords to the at least one domain (also referred to as topic) may be initiated. Initially, at step 704, a semantic clustering operation may be performed on the plurality of entry-level tasks. In other words, the plurality of entry-level tasks may be processed to generate the plurality of semantic clusters. In an embodiment, to generate the plurality of semantic clusters, the predefined clustering technique, e.g., the K-means clustering may be applied.
[080] Once the plurality of semantic clusters are generated, at step 706, a plurality of keywords may be identified within each cluster. In an embodiment, the plurality of keywords may be identified using keyword extraction techniques such as Term Frequency – Inverse Document Frequency (TF-IDF) scoring, embedding-based keyword selection, part-of-speech filtering, attention-based relevance scoring, etc. The plurality of keywords may represent salient concepts, domains, or task attributes that are characteristic of each corresponding cluster. Once the plurality of keywords is identified, at step 708, each of the plurality of keywords may be mapped to the at least one domain (i.e., the topic). In some embodiments, a domain-mapping table, ontology, keyword-domain dictionary, or machine-learned classifier may be used to associate each keyword with the at least one domain. Further, the process 700 may terminate at step 710.
[081] Referring now to FIG. 8, a detailed flow diagram of a process 800 for arranging the plurality of semantic clusters is illustrated, in accordance with some embodiments of the present disclosure. FIG. 8 is explained in conjunction with FIGS. 1, 2, 3, 4, 5, 6, and 7. Further, each step of the method 800 may be implemented by various modules present within the memory 104 of the computing device 102 of the system 100.
[082] At step 802, the process 800 for arranging the plurality of semantic clusters in the sequential learning order may be initiated. At step 804, the plurality of semantic clusters along with the number of topics (i.e., the number of domains) associated with each cluster may be obtained. Further, an empty ordered list ‘L’ may be initialized to hold each of the plurality of semantic clusters in the sequential learning order. To arrange the plurality of semantic clusters, at step 806, a cluster, e.g., a cluster C (i.e., the first cluster), having a minimum number of topics may be selected. In an embodiment, in an event of a tie between two or more clusters having the same number of minimum topic count, the tie between the two or more clusters may be broken arbitrarily or according to a predefined tie-breaking rule (for example, by cluster identifier, creation timestamp, or a secondary ranking metric).
[083] Further, at step 808, the selected cluster (e.g., the cluster C) may be inserted into the ordered list L. Thereafter, at step 810, a check may be performed to determine whether all the plurality of semantic clusters have been inserted into the ordered list L. In one embodiment, based on the check performed at step 810, if all the plurality of semantic clusters are inserted into the ordered list L, the process 800 may terminate at step 812. In another embodiment, based on the check performed at step 810, if all the plurality of semantic clusters are not inserted into the ordered list L, step 814 is executed. At step 814, the subsequent cluster (e.g., a next cluster C) having a minimum number of new topics (i.e., topics not previously introduced by any of the clusters already inserted into list L) may be identified. Upon identifying the subsequent cluster, steps 808, 810, and 814 may be re-executed iteratively until all the plurality of semantic clusters are inserted into the ordered list L.
[084] Referring now to FIG. 9, a detailed flow diagram of a process 900 for monitoring the performance of the user is illustrated, in accordance with some embodiments of the present disclosure. FIG. 9 is explained in conjunction with FIGS. 1, 2, 3, 4, 5, 6, 7, and 8. Further, each step of the method 900 may be implemented by various modules present within the memory 104 of the computing device 102 of the system 100.
[085] At step 902, the process 900 for monitoring the performance of the user may be initiated. At step 904, one or more variables, e.g., the difficulty level for each of the plurality of semantic clusters may be initialized. For example, an initial value for the difficulty level of each of the plurality of semantic clusters may be set for average. Further at step 906, a check may be performed to determine whether all the plurality of semantic clusters have been covered for the current user during the training process. In one embodiment, based on the check performed at step 906, if all the plurality of semantic clusters are determined to have been covered, the process 900 terminates at step 908. In another embodiment, based on the check performed at step 906, if one or more clusters of the plurality of semantic clusters are determined to remain uncovered, step 910 may be executed. At step 910, the next cluster, i.e., the subsequent cluster may be selected according to the sequential learning order.
[086] Upon selecting the next cluster, at step 912, a time budget (i.e., the training time) may be allocated for the selected cluster (represented as cluster time limit) and initializes a counter depicting an actual cluster time taken to zero. In addition, at step 912, the one or more contextual learning resources may be provided to the user for any new topics introduced by the selected cluster based on the predefined trigger. Once the training time is allocated, at step 914, a current assignment may be choosen from the set of assignments associated with the selected cluster having the difficulty level as below average, average, or above average. With reference to FIG. 2, each assignment may be performed by the user via the study environment module 208. Further, at step 916, the one or more performance parameters, e.g., the total marks obtained by the user and the total time taken by the user may be determined. Further, based on the one or more performance parameters, at step 918, a check may be performed to determine whether the total marks obtained by the user for the current assignment is below or above minimum marks. In an embodiment, the minimum marks to be obtained for each assignment may be specified by a course instructor, a subject matter expert, or the like. In one embodiment, based on the check performed at step 918, if the total marks obtained by the user is above the minimum marks, the performance of the user may be determined to be above average and the step 920 may be executed. At step 920, based on determining the performance of the user to be above average, the difficulty level for assignments may be increased. For example, if the user was performing assignments with average difficulty level, then the difficulty level of subsequent assignments may be inreased to above average. Upon increasing the difficulty level for subsequent assignments, step 924 may be executed.
[087] In another embodiment, based on the check performed at step 918, if the total marks obtained by the user is below the minimum marks, the performance of the user may be determined to be below average and the step 922 may be executed. At step 922, based on determining the performance of the user to be below average, the difficulty level for assignments may be decreased. For example, if the user was performing assignments with average difficulty level, then the difficulty level of subsequent assignments may be decreased to below average. Upon decreasing the difficulty level for subsequent assignments, step 924 may be executed.
[088] Further, at step 924, the total time taken by the user to complete each of the set of assignments within the current cluster may be updated to include a most recent timestamp, depicted as an actual cluster time taken is equal to an actual cluster time taken (i.e., a cumulative total time that the user has spent on all previous assignments within the current cluster) + time taken (i.e., a time duration the user spent on the most recently completed assignment within the current cluster.). Once the total time taken (i.e, the actual cluster time taken) by the user is updated, at step 926, a check is performed to determine if the total time taken by the user to complete each of the set of assignments within the current cluster exceeds the allocated training time (depicted as cluster time limit) for the current cluster. In an embodiment, based on the check performed at step 926, steps 906-926 may be re-executed. In one embodiment, if the total time does not exceed the allocated training time, steps 914-926 may be re-executed for the same cluster. In another embodiment, if the total time exceeds the allocated training time, steps 906-926 may be re-executed for the next cluster.
[089] In an embodiment, the disclosed invention introduces several key novelties that may collectively enable an improved and technically robust framework for training human agents (i.e., the user or the trainee) using agentic AI–derived knowledge. Firstly, the disclosed invention may provide an adaptive methodology that leverages insights generated by agentic AI systems (i.e., the LLM) to structure and personalize human learning pathways. Secondly, the disclosed invention may incorporate a weighted scoring mechanism for objectively quantifying and ranking a complexity of problems (i.e., entry-level tasks) solved by the LLM, thereby enabling consistent assessment of task difficulty across diverse domains. Thirdly, the disclosed invention may present a unique method for arranging the entry-level task in a progressive order, i.e., the sequential learning order, by transitioning from tasks that require lower skill levels to those demanding higher proficiency and from tasks of lesser complexity to those of greater complexity, thereby ensuring that the user follows an optimized learning trajectory. Additionally, the disclosed invention may introduce an adaptive task-presentation strategy that dynamically selects and delivers tasks to the users based on their demonstrated performance and evolving skill profile, resulting in a continuously calibrated and personalized training experience. As a result, the present invention may generate a highly personalized and effective training curriculum that balances skill acquisition with practical application and is deployable in both cloud-based and local network environments.
[090] As will be also appreciated, the above-described techniques may take the form of computer or controller implemented processes and apparatuses for practicing those processes. The disclosure can also be embodied in the form of computer program code containing instructions embodied in tangible media, such as floppy diskettes, solid state drives, CD-ROMs, hard drives, or any other computer-readable storage medium, wherein, when the computer program code is loaded into and executed by a computer or controller, the computer becomes an apparatus for practicing the invention. The disclosure may also be embodied in the form of computer program code or signal, for example, whether stored in a storage medium, loaded into and/or executed by a computer or controller, or transmitted over some transmission medium, such as over electrical wiring or cabling, through fiber optics, or via electromagnetic radiation, wherein, when the computer program code is loaded into and executed by a computer, the computer becomes an apparatus for practicing the invention. When implemented on a general-purpose microprocessor, the computer program code segments configure the microprocessor to create specific logic circuits.
[091] The disclosed methods and systems may be implemented on a conventional or a general-purpose computer system, such as a personal computer (PC) or server computer. Referring now to FIG. 10, an exemplary computing system 1000 that may be employed to implement processing functionality for various embodiments (e.g., as a Single Instruction, Multiple Data (SIMD) device, client device, server device, one or more processors, or the like) is illustrated. Those skilled in the relevant art will also recognize how to implement the invention using other computer systems or architectures. The computing system 1000 may represent, for example, a user device such as a desktop, a laptop, a mobile phone, personal entertainment device, Digital Video Recorder (DVR), and so on, or any other type of special or general-purpose computing device as may be desirable or appropriate for a given application or environment. The computing system 1000 may include one or more processors, such as a processor 1002 that may be implemented using a general or special purpose processing engine such as, for example, a microprocessor, microcontroller or other control logic. In this example, the processor 1002 is connected to a bus 1004 or other communication medium. In some embodiments, the processor 1002 may be an Artificial Intelligence (AI) processor, which may be implemented as a Tensor Processing Unit (TPU), or a graphical processor unit, or a custom programmable solution Field-Programmable Gate Array (FPGA).
[092] The computing system 1000 may also include a memory 1006 (main memory), for example, Random Access Memory (RAM) or other dynamic memory, for storing information and instructions to be executed by the processor 1002. The memory 1006 also may be used for storing temporary variables or other intermediate information during execution of instructions to be executed by the processor 1002. The computing system 1000 may likewise include a Read Only Memory (ROM) or other static storage device coupled to bus 1004 for storing static information and instructions for the processor 1002.
[093] The computing system 1000 may also include storage devices 1008, which may include, for example, a media drive 1010 and a removable storage interface. The media drive 1010 may include a drive or other mechanism to support fixed or removable storage media, such as a hard disk drive, a floppy disk drive, a magnetic tape drive, an Secure Digital (SD) card port, a Universal Serial Bus (USB) port, a micro-USB, an optical disk drive, a Compact Disc (CD) or Digital Versatile Disc (DVD) drive (R or Rewritable (RW)), or other removable or fixed media drive. A storage media 1012 may include, for example, a hard disk, magnetic tape, flash drive, or other fixed or removable medium that is read by and written to by the media drive 1010. As these examples illustrate, the storage media 1012 may include a computer-readable storage medium having stored therein particular computer software or data.
[094] In alternative embodiments, the storage devices 1008 may include other similar instrumentalities for allowing computer programs or other instructions or data to be loaded into the computing system 1000. Such instrumentalities may include, for example, a removable storage unit 1014 and a storage unit interface 1016, such as a program cartridge and cartridge interface, a removable memory (for example, a flash memory or other removable memory module) and memory slot, and other removable storage units and interfaces that allow software and data to be transferred from the removable storage unit 1014 to the computing system 1000.
[095] The computing system 1000 may also include a communications interface 1018. The communications interface 1018 may be used to allow software and data to be transferred between the computing system 1000 and external devices. Examples of the communications interface 1018 may include a network interface (such as an Ethernet or other Network Interface Card (NIC) card), a communications port (such as for example, a USB port, a micro-USB port), Near field Communication (NFC), etc. Software and data transferred via the communications interface 1018 are in the form of signals which may be electronic, electromagnetic, optical, or other signals capable of being received by the communications interface 1018. These signals are provided to the communications interface 1018 via a channel 1020. The channel 1020 may carry signals and may be implemented using a wireless medium, wire or cable, fiber optics, or other communications medium. Some examples of the channel 1020 may include a phone line, a cellular phone link, an RF link, a Bluetooth link, a network interface, a local or wide area network, and other communications channels.
[096] The computing system 1000 may further include Input/Output (I/O) devices 1022. Examples may include, but are not limited to a display, keypad, microphone, audio speakers, vibrating motor, Light Emitting Diode (LED) lights, etc. The I/O devices 1022 may receive input from a user and also display an output of the computation performed by the processor 1002. In this document, the terms “computer program product” and “computer-readable medium” may be used generally to refer to media such as, for example, the memory 1006, the storage devices 1008, the removable storage unit 1014, or signal(s) on the channel 1020. These and other forms of computer-readable media may be involved in providing one or more sequences of one or more instructions to the processor 1002 for execution. Such instructions, generally referred to as “computer program code” (which may be grouped in the form of computer programs or other groupings), when executed, enable the computing system 1000 to perform features or functions of embodiments of the present invention.
[097] In an embodiment where the elements are implemented using software, the software may be stored in a computer-readable medium and loaded into the computing system 1000 using, for example, the removable storage unit 1014, the media drive 1010 or the communications interface 1018. The control logic (in this example, software instructions or computer program code), when executed by the processor 1002, causes the processor 1002 to perform the functions of the invention as described herein.
[098] Various embodiments provide a method and a system for automatically generating a personalized training curriculum for a user. The disclosed method and system may receive information corresponding to a plurality of entry-level tasks executed on a job from a Large Language Model (LLM) at a predefined time interval. The disclosed method and system may process the information corresponding to the plurality of entry-level tasks to determine a difficulty score for each of the plurality of entry-level tasks based on a set of parameters. The disclosed method and system may generate a plurality of semantic clusters including one or more of the plurality of entry-level tasks, based on the difficulty score determined for each entry-level task, using a pre-defined clustering technique. The disclosed method and system may arrange each of the plurality of semantic clusters in a sequential learning order. The disclosed method and system may generate the personalized training curriculum including a set of assignments for the user based on the sequential learning order of each of the plurality of semantic clusters. The disclosed method and system may render the personalized training curriculum via a Graphical User Interface (GUI) to the user.
[099] Thus, the disclosed method and system may try to overcome the technical problem of automatically generating a personalized training curriculum for a user. The disclosed method and system may enable Artificial Intelligence (AI) - driven workforce development through a multi-agent framework (i.e., the computing device 102) that bridges the gap between automated tasks and human skill acquisition, thereby ensuring a continuously competent and future-ready workforce. The disclosed method and system may facilitate proactive skill development by anticipating skill gaps created by generative AI (GenAI) automation and autonomously training new hires for more complex tasks (also referred to a L2-level task), ensuring a sustainable and skilled talent pipeline. The disclosed method and system may further support personalized and adaptive training by dynamically adjusting the personalized training curriculum based on progress, performance, and available time for each user (e.g., a trainee), thereby avoiding a one-size-fits-all approach and promoting efficient and effective learning outcomes.
[0100] Additionally, the disclosed method and system may provide automated and scalable onboarding by streamlining and digitizing traditionally manual training processes, allowing organizations to onboard and train large cohorts of new hires simultaneously while reducing administrative overhead and time-to-productivity. The disclosed method and system may also maximize human potential by allowing employees’ to upskill quickly from repetitive entry level taks (i.e., L1 tasks) to higher level tasks (i.e., L2-level tasks) responsibilities, enabling greater workforce contribution, creativity, and engagement.
[0101] Moreover, the disclosed method and system may enhance organizational agility and resilience by continuously aligning workforce (i.e., employees) skills with evolving business needs and emerging technologies, ensuring operational readiness in dynamic environments. The disclosed method and system may further improve cost and resource efficiency by reducing dependence on manual training interventions and optimizing learning cycles, allowing Human Resource (HR) and L&D (Learning and Development) teams to focus on strategic workforce planning. In addition, the disclosed method and system may promote a future-ready enterprise, capable of sustaining innovation and competitiveness through continuous, AI-assisted employee development and upskilling.
[0102] In light of the above-mentioned advantages and the technical advancements provided by the disclosed method and system, the claimed steps as discussed above are not routine, conventional, or well understood in the art, as the claimed steps enable the following solutions to the existing problems in conventional technologies. Further, the claimed steps clearly bring an improvement in the functioning of the device itself as the claimed steps provide a technical solution to a technical problem.
[0103] The specification has described a method and system for generating a personalized training curriculum for a user. The illustrated steps are set out to explain the exemplary embodiments shown, and it should be anticipated that ongoing technological development will change the manner in which particular functions are performed. These examples are presented herein for purposes of illustration, and not limitation. Further, the boundaries of the functional building blocks have been arbitrarily defined herein for the convenience of the description. Alternative boundaries can be defined so long as the specified functions and relationships thereof are appropriately performed. Alternatives (including equivalents, extensions, variations, deviations, etc., of those described herein) will be apparent to persons skilled in the relevant art(s) based on the teachings contained herein. Such alternatives fall within the scope and spirit of the disclosed embodiments.
[0104] Furthermore, one or more computer-readable storage media may be utilized in implementing embodiments consistent with the present disclosure. A computer-readable storage medium refers to any type of physical memory on which information or data readable by a processor may be stored. Thus, a computer-readable storage medium may store instructions for execution by one or more processors, including instructions for causing the processor(s) to perform steps or stages consistent with the embodiments described herein. The term “computer-readable medium” should be understood to include tangible items and exclude carrier waves and transient signals, i.e., be non-transitory. Examples include random access memory (RAM), read-only memory (ROM), volatile memory, nonvolatile memory, hard drives, CD ROMs, DVDs, flash drives, disks, and any other known physical storage media.
[0105] It is intended that the disclosure and examples be considered as exemplary only, with a true scope and spirit of disclosed embodiments being indicated by the following claims. , Claims:CLAIMS
WHAT IS CLAIMED IS:
1. A method (300) for generating a personalized training curriculum for a user, the method comprising:
receiving (302), by a processor (106), information corresponding to a plurality of entry-level tasks executed on a job from a Large Language Model (LLM) at a predefined time interval;
processing (304), by the processor (106), the information corresponding to the plurality of entry-level tasks to determine a difficulty score for each of the plurality of entry-level tasks based on a set of parameters;
generating (306), by the processor (106), a plurality of semantic clusters comprising one or more of the plurality of entry-level tasks, based on the difficulty score determined for each entry-level task, using a pre-defined clustering technique;
arranging (308), by the processor (106), each of the plurality of semantic clusters in a sequential learning order;
generating (310), by the processor (106), the personalized training curriculum comprising a set of assignments for the user based on the sequential learning order of each of the plurality of semantic clusters; and
rendering (312), by the processor (106), the personalized training curriculum via a Graphical User Interface (GUI) to the user.

2. The method (300) as claimed in claim 1, wherein the information corresponding to the plurality of entry-level tasks comprises a domain associated with each task, a solution associated with each task, an estimated completion time for each task, a LLM confidence score for the solution associated with each task, and a set of resources required for each task.

3. The method (300) as claimed in claim 1, wherein the set of parameters comprises an expected time to complete each task, a number of questions historically asked to complete each task, and the LLM confidence score for the solution associated with each task.

4. The method (300) as claimed in claim 1, wherein arranging (308) each of the plurality of semantic clusters in the sequential learning order comprises:
selecting (402), by the processor (106), a first cluster with a minimum number of domains, from the plurality of semantic clusters;
iteratively selecting (404), by the processor (106), a subsequent cluster with a minimum number of new domains not previously introduced in the first cluster; and
repeating (406), by the processor (106), the selection until each of the plurality of semantic clusters are arranged in the sequential learning order.

5. The method (300) as claimed in claim 4, wherein a number of domains for each cluster of the plurality of semantic clusters is determined by identifying one or more keywords from each cluster and mapping each of the one or more keywords to at least one domain.

6. The method (300) as claimed in claim 5, comprising:
allocating, by the processor (106), a training time to each cluster based on the number of domains introduced by a corresponding cluster.

7. The method (300) as claimed in claim 1, wherein rendering (312) the personalized training curriculum to the user comprising:
rendering (502), by the processor (106) via the GUI, an assignment from the set of assignments within the personalized training curriculum to the user.

8. The method (300) as claimed in claim 7, comprising:
receiving (504), by the processor (106) via the GUI, a user response corresponding to a set of tasks provided in the assignment;
monitoring in real-time (506), by the processor (106), a performance of the user based on the received user response and one or more performance parameters; and
dynamically adjusting (508), by the processor (106), a difficulty level of a subsequent assignment to be rendered to the user based on the performance of the user.

9. The method (300) as claimed in claim 8, wherein the one or more performance parameters comprises total marks obtained by the user for the assignment, a total time taken by the user to complete the assignment, and a completion status of the assignment by the user.

10. The method (300) as claimed in claim 8, comprising:
 identifying, by the processor (106), one or more contextual learning resources based on the performance of the user and the nature of a current task; and
displaying, by the processor (106) via the GUI, the one or more contextual learning resources to the user at a predefined trigger during an execution of the current task.

11. A system (100) for automatically generating a personalized training curriculum for a user, the system comprising:
a processor (106); and
a memory (104) coupled to the processor (106), wherein the memory (104) stores processor executable instructions, which, on execution, causes the processor (106) to:
receive (302) information corresponding to a plurality of entry-level tasks executed on a job from a Large Language Model (LLM) at a predefined time interval;
process (304) the information corresponding to the plurality of entry-level tasks to determine a difficulty score for each of the plurality of entry-level tasks based on a set of parameters;
generate (306) a plurality of semantic clusters comprising one or more of the plurality of entry-level tasks, based on the difficulty score determined for each entry-level task, using a pre-defined clustering technique;
arrange (308) each of the plurality of semantic clusters in a sequential learning order;
generate (310) the personalized training curriculum comprising a set of assignments for the user based on the sequential learning order of each of the plurality of semantic clusters; and
render (312) the personalized training curriculum via a Graphical User Interface (GUI) to the user.

12. The system (100) as claimed in claim 11, wherein the information corresponding to the plurality of entry-level tasks comprises a domain associated with each task, a solution associated with each task, an estimated completion time for each task, a LLM confidence score for the solution associated with each task, and a set of resources required for each task.

13. The system (100) as claimed in claim 11, wherein the set of parameters comprises an expected time to complete each task, a number of questions historically asked to complete each task, and the LLM confidence score for the solution associated with each task.

14. The system (100) as claimed in claim 11, wherein, to arrange (308) each of the plurality of semantic clusters in the sequential learning order, the processor executable instructions cause the processor (106) to:
select (402) a first cluster with a minimum number of domains, from the plurality of semantic clusters;
iteratively select (404) a subsequent cluster with a minimum number of new domains not previously introduced in the first cluster; and
repeat (406) the selection until each of the plurality of semantic clusters are arranged in the sequential learning order.

15. The system (100) as claimed in claim 14, wherein a number of domains for each cluster of the plurality of semantic clusters is determined by identifying one or more keywords from each cluster and mapping each of the one or more keywords to at least one domain.

16. The system (100) as claimed in claim 15, wherein the processor executable instructions cause the processor (106) to:
allocate a training time to each cluster based on the number of domains introduced by a corresponding cluster.

17. The system (100) as claimed in claim 11, wherein, to render (312) the personalized training curriculum to the user, the processor executable instructions cause the processor (106) to:
render (502) via the GUI, an assignment from the set of assignments within the personalized training curriculum to the user.

18. The system (100) as claimed in claim 17, wherein the processor executable instructions cause the processor (106) to:
receive (504) via the GUI, a user response corresponding to a set of tasks provided in the assignment;
monitor in real-time (506), a performance of the user based on the received user response and one or more performance parameters; and
dynamically adjust (508) a difficulty level of a subsequent assignment to be rendered to the user based on the performance of the user.

19. The system (100) as claimed in claim 18, wherein the one or more performance parameters comprises total marks obtained by the user for the assignment, a total time taken by the user to complete the assignment, and a completion status of the assignment by the user.

20. The system (100) as claimed in claim 18, wherein the processor executable instructions cause the processor (106) to:
identify one or more contextual learning resources based on the performance of the user and the nature of a current task; and
display via the GUI, the one or more contextual learning resources to the user at a predefined trigger during an execution of the current task.