Bridging Mobile Development and Scientific Machine Learning, with a growing focus on renewable energy and climate tech.
🩺 Euexia — Android Health Tracker
A production-grade health app, built solo — React + Vite + Capacitor with a custom native Java foreground service for true background step counting, Supabase cloud sync, and OTA updates.
Private source · public portfolio showcase · Play Store launch in progress.
- 🔐 SecureVault — a live DRM & software-licensing platform (demo) (FastAPI · Next.js · PostgreSQL · Razorpay — deployed on Render + Neon + Vercel)
- 🤖 Alfred — a multi-agent AI assistant deployed on the edge (live demo)
- ☀️ PhysSolar — evolving my hybrid physics + ML solar forecaster toward a Physics-Informed Neural Network
- 📐 Deepening Mathematics & Physics for Scientific ML
Frontend & Mobile
Backend & Full-Stack
AI & Data Science
LLMs & AI Engineering
Tools & Platforms
- 🩺 Engineered a native Android foreground service (Java) for reliable background step counting — the engine behind Euexia.
- 🔬 PhysSolar beats clear-sky and persistence baselines (0.85 / 0.69 skill) over a full year of real hourly data, with a leakage-free forward-chaining backtest.
- 🤖 Deployed Alfred, a multi-agent AI system, to the edge — streaming RAG chat, persistent memory, and a red-team-tested abuse guard.
- 🔐 Built SecureVault end to end — JWT + TOTP auth, Razorpay payments with HMAC-verified webhooks, database migrations, and a comprehensive passing test suite.
A structured 18–24 month path into Scientific Machine Learning — pairing deep technical skills with real domain knowledge in physics and clean energy.
- Build a physics-informed solar forecasting model — PhysSolar
- Master the mathematics for ML
- Deploy a Physics-Informed Neural Network (PINN)
οὐ μόνον μαθὼν ἀλλὰ καὶ παθών τὰ θεῖα

