An Ai-driven Task Management System With Role-based Access Control and Llm-assisted Agents: Architecture, Implementation, and Evaluation
Year: 2026
Type: article
Abstract: Planning software projects involves repeated manual decisions: decomposing a project into tasks, deciding who should perform them, tracking progress, and noticing when workloads become unbalanced.This paper presents the design and implementation of an AI-Driven Task Management System that supports these decisions through a web application backed by large language models (LLMs).The system consists of a React single-page client, a FastAPI backend, and a PostgreSQL database, with access governed by JWT authentication and four organizational roles.LLMassisted agent functions served through the Groq API decompose a project into tasks and assign them to team members, suggest project teams, summarize meeting transcripts, generate project and member reports, and analyze workload and burnout risk.Model output requested as JSON is parsed and matched against the stored team roster, and a deterministic skill-and-workload scoring rule assigns the task when the Ramanandhini K, Manoranjitham R, Girivasan K, Sakthivel L, Dakshin Kumar M https://iaeme.com/Home/journal/IJAIML277 [email protected]'s choice cannot be matched.The paper documents the architecture and algorithms as implemented in the repository and defines an evaluation protocol.
Language: English
Subfield: Information Systems
Field: Computer Science
Domain: Physical Sciences
Open Access status: gold
OpenAlex ID: W7230651903
