Full-stack and AI developer in Chennai, building offline-first and explainable AI apps.
I like software that still works with no network, and models that can show their reasoning.
- Offline-first apps that keep working on slow or missing networks (edge AI for classrooms, local-only Windows tools).
- Explainable AI: heatmaps, marked regions and plain-language reports instead of a bare score.
- Cross-platform products in Flutter and Next.js with a Python backend.
- Local LLMs on Ollama, so data stays on the machine.
Languages
Frameworks and apps
AI and data
Backend and cloud
Flask + PyTorch research prototype that classifies scan type, draws Hi-Res CAM heatmaps with marked regions, and exports multilingual PDF reports, with a Flutter companion app for patients and doctors. Not for clinical use.
Python PyTorch Flask Flutter
🖱️ DropWheel
Hold Shift, Ctrl or Alt while dragging a file in Windows Explorer and a wheel appears: drop on it to convert, compress, resize, crop, merge, split or protect images, PDFs, video and audio. Fully offline, built on open-source engines.
C# Windows Offline
Edge-first AI diagnostic and remediation copilot for foundational learning (FLN / NIPUN Bharat), designed to run where connectivity is poor.
JavaScript Edge AI EdTech
- 📺 tv-toolkit: read-first, reversible tools for Android TV: debloat, boot speed, privacy audit, launcher, Dolby pass-through player.
- 🌍 AI-World: a 3D globe where ten tiny neural networks choose what to learn, mentored by a local LLM on Ollama.
- 🍛 Canteen-Food-App: Flutter + Supabase pre-ordering app with student and shop-owner roles.
I'm open to full-stack, Flutter and applied-AI roles, and happy to review issues or PRs on any repo above.



