RAG system using LangGraph multi-agent workflows with Azure OpenAI and persistent ChromaDB storage. Ingests and chunks multi-format documents, performs semantic retrieval with source attribution, and streams grounded answers. Includes a retrieval inspector agent that evaluates chunk ranking, recall quality, and answer groundedness.

Autonomous research agent with a full Planner → Researcher → Summarizer → Critic cycle, retrieving Tavily web and arXiv sources, drafting and improving cited reports using Azure-hosted LLMs, with ChromaDB-backed long-term memory for cross-run knowledge persistence enabling iterative, self-improving research workflows and reusable knowledge.

Network Security tutor built with a Django REST API and Next.js frontend, using an Ollama-hosted model for streaming tutor chat, adaptive quiz generation, and grading. Uses ChromaDB with sentence-transformer embeddings to support a personalized knowledge base, persistent sessions, and context-aware learning across textbooks and lecture slides.

Full-stack job application tracker built with FastAPI, PostgreSQL, and a Next.js + TypeScript frontend. Supports stage-based workflow management, structured interview round tracking, resume upload with S3-backed storage and in-app preview, and flow analytics to visualize stage transitions from application to final outcome.

Co-authored peer-reviewed paper, “Generative Adversarial Networks for Data Augmentation in Image Recognition: An Exploratory Study” (IJAIML, 2025), exploring GAN-generated synthetic data to improve image classification under limited labeled data. Demonstrated measurable accuracy gains by integrating realistic synthetic images to strengthen model generalization.