AI/ML Engineer & Full-Stack Developer · IT undergrad @ BVM Engineering College (2027)
I build intelligent systems that bridge research and real-world impact.
- 🔬 AI/ML research intern at University of Wollongong, GIFT City (May–Jun 2026)
- 🏆 MatchDay Copilot placed 117 / 43,000 at PromptWars — a grounded-GenAI stadium assistant
- 🩺 Currently building a multi-label retinal disease classifier — segmentation masks as auxiliary input channels for glaucoma, hypertensive and diabetic retinopathy detection
- 🛰️ Into geospatial ML — satellite-derived features (NDVI, nighttime lights, built-up density) for urban growth forecasting
- 🌐 Open-source contributor, GSSoC 2026 · Tech Core Member, GDGC BVM
- 💬 Ask me about grounded LLM pipelines, model deployment, or React front-ends
Languages
ML / AI
Web
| Project | What it does | Stack |
|---|---|---|
MatchDay Copilot · 117/43k |
GenAI matchday assistant for FIFA WC 2026 venues — Dijkstra in-stadium wayfinding with a guaranteed step-free mode, live crowd advisories, and a Claude tier grounded in a per-request context pack so it can't invent gate numbers. Falls back to a deterministic offline router with zero connectivity. | FastAPI, Claude API, Dijkstra, pytest |
| UrbanGrowth | Forecasts urban expansion in Gandhinagar to 2040 from satellite-derived signals — built-up %, NDVI, nighttime lights — trained on 2016–2024 history and served over an interactive Leaflet map with area comparison. | scikit-learn, FastAPI, Gemini, React, Leaflet |
| Retinal Disease Classifier | Multi-label detection of glaucoma, hypertensive and diabetic retinopathy — segmentation masks fed in as auxiliary input channels rather than a separate multi-task head. | PyTorch, OpenCV |
| AtEase | Campus mental health & wellbeing platform for students. | React, Node, Express |