AI/ML & full-stack engineer. I build end-to-end products (Django,FastAPI, React, Flutter) and spend my spare time on mechanistic interpretability research reverse-engineering what's actually happening inside transformers rather than just calling them black boxes.
🎯 Currently looking for full-time SWE / ML engineering roles.
Recent: merged a fix into docling-project/docling (IBM's document-AI toolkit) corrected a dehyphenation bug in the PDF text-extraction pipeline that was silently corrupting CLI flags in technical docs. Shipped in v2.122.0.
| Project | What it does |
|---|---|
| MediLens | AI medical assistant — Gemini-driven symptom triage/NER + local PubMed RAG + prescription OCR |
| FIDES | Multi-stage AI pipeline for trade-document fraud/compliance risk screening |
| Fraud Detection | Real-time fraud detection under extreme class imbalance — XGBoost/Isolation Forest/autoencoder, served via FastAPI |
| Diagnostic Assistant | Medical imaging diagnosis with Grad-CAM explainability |
| Driftmesh | Serverless P2P mesh chat over Bluetooth LE, Rust core + Flutter UI, E2E encrypted |
| BrainWave | Full-stack ed-tech platform — course management, OTP auth, Razorpay payments |
Python TypeScript/JavaScript C++ Rust FastAPI React Flutter PyTorch Django
