AI engineer @ Capgemini — building retrieval and agent systems on top of LLMs. Mostly Python, FastAPI, and the Azure AI stack. Previously: B.Tech in IT, GL Bajaj Institute.
A mix of shipped work and honest work-in-progress — these are the repos I'd point at if you asked what I've built.
- MiA-RAG — paper-accurate implementation of Mindscape-Aware RAG (arXiv:2512.17220) with the official MiA-Emb-0.6B embedding model
- llm-learning-log — daily notes from working through Hands-On LLMs
- ahad.works — self-hosted infra: Caddy, Docker, deploy scripts (infra)
- ai-gym-memory-system — conversational logger that turns free-text workout and activity input into structured records, with Gemini intent extraction and embedding-based semantic recall
- document-intelligence-rag — enterprise RAG on Azure AI Search: chunking, vector embeddings, grounded generation, with hallucination checks
- Agentic-RAG — multi-agent RAG for manufacturing document Q&A
- MiA-RAG — research-paper reproduction (see Now)
- workspace-mirror-skill — an agent skill for the Antigravity IDE that mirrors conversation and IDE state into a standalone, offline HTML dashboard
Python · FastAPI · AG2 · LangChain · Azure AI Search / OpenAI · ChromaDB · Docker · Java