Final-year CSE (AI/ML) engineer building where machine learning meets infrastructure — in-network intelligence, behavioral security, and retrieval-augmented systems.
Systems where ML has to survive contact with real infrastructure constraints — not just notebooks.
| Domain | Focus |
|---|---|
| 🕸️ In-Network ML | Programmable data planes (P4), gradient aggregation at line-rate |
| 🛡️ Behavioral Security | Keystroke/mouse dynamics, continuous auth, anomaly detection |
| 📚 Applied RAG | Retrieval systems built from primitives, not frameworks |
| 🌐 Federated Learning | Distributed training with network-layer optimization |
Full-stack RAG app generating cited research drafts with per-claim trust scoring.
FastAPI React/Vite Groq Llama-3 Supabase pgvector
Built retrieval logic from scratch (no LangChain) — graceful degradation to TF-IDF under memory constraints on free-tier hosting.
Cognitive identity verification via behavioral biometrics — real-time intruder detection with automated countermeasures.
TypeScript
Gradient summation performed inside programmable switches — cuts network congestion and server overhead.
P4
Line-rate volumetric SYN flood detection using P4/BMv2 — zero CPU overhead.
Python P4 BMv2
CNN on Mel-spectrograms for industrial predictive maintenance, deployed via Streamlit + Docker.
Python TensorFlow
Languages
ML / AI
Web & Backend
Networking & Security
Tools
