Multi-Agent System for Secure Code Generation and Vulnerability Remediation

Key Technical Highlights:

• Multi-Agent System (MAS): Designed and implemented a collaborative MAS using LangGraph, where specialized AI agents (OWASP ASVS-aligned) analyze and secure code.
• AI-Driven Analysis & Generation: Leverages LLMs for contextual code understanding, secure code generation, and vulnerability remediation, integrated with SAST (Semgrep) and Tree-sitter.
• Full-Stack Development: Backend built with Python (FastAPI, LangGraph), asynchronous task processing with RabbitMQ & Celery-like worker pattern. Frontend developed using React and TypeScript.
• DevSecOps Workflow: Demonstrates an automated end-to-end pipeline from code submission, multi-path analysis, fixing suggestions, to detailed reporting.
• Dynamic Code Parsing: Utilizes Tree-sitter for robust, multi-language AST generation and querying, enabling precise code analysis.
• User Authentication & API Security: Implemented JWT-based authentication using `fastapi-users` and designed secure API endpoints.