Applied AI & Full-Stack Developer
I build practical AI systems, research-focused ML pipelines, and full-stack products that turn complex problems into usable software. My work sits at the intersection of retrieval systems, computer vision, cybersecurity research, and backend-heavy product engineering.
ship the experiment. test the story. improve the system.
I'm interested in building AI that is not just impressive but measurable, grounded, and useful in real workflows. I like working across the stack: shaping ML experiments, designing APIs, building product interfaces, and making sure the system actually survives real inputs.
Currently, I'm focused on ST-HF for video visible-infrared person re-identification. Next, I'm continuing work on Veridian AI, an evaluation-driven enterprise query engine for grounded analytical answers.
| Project | What it does | Stack |
|---|---|---|
| ST-HF VVI-ReID | Video visible-infrared person re-identification work | PyTorch, ResNet-50, Computer Vision |
| Veridian AI | Enterprise query engine for grounded analytical answers | Python, FastAPI, DuckDB, RAG |
| ToN-IoT IDS | Rare-class MITM attack detection on IoT network traffic | LightGBM, KMeansSMOTE, XGBoost |
| Astra-Q | KG + RAG help bot for ISRO's MOSDAC workflows | Neo4j, LangChain, FAISS, Gemini |
| Vanta AI | Deepfake detection platform and digital safety product | React, Node.js, FastAPI |
- Working on ST-HF VVI-ReID for video visible-infrared person re-identification
- Building Veridian AI as an evaluation-driven enterprise query engine
- Exploring rare-class intrusion detection for IoT network traffic
- Connecting research prototypes to clean APIs and usable product interfaces
