📍 Pune, India · ⚙️ Backend + applied AI · 🎓 Final-year Computer Engineering student
I like building software that holds up in the real world, then making it cheaper, faster and harder to break.
- 🧾 Concord: a B2B invoice dispute platform that reconciles POs, delivery notes and invoices, flags mismatches by severity, and manages vendor disputes through to settlement.
- 🚦 Release Captain: a multi-agent system that reviews pull requests in parallel and posts a ship / hold / rollback verdict to GitHub and Slack.
- 🏪 Udhaar: a local AI back-office for kirana shops that runs on the shop's own machine (Hacktoberfest 2026, Week 2).
- 🛡️ Kavach: a scam shield for older adults on Alexa+, backed by a self-hosted MCP server (Amazon Developer Hackathon 2026).
Earlier builds
- 🗺️ SmritiPath: turns recorded spoken memories into a walking tour on a map, all local AI (Hacktoberfest 2026, Week 1).
- 🩺 Health Assistant: AI health information assistant (PromptWars one-day hackathon).
- 🛒 MakhanaMagic: e-commerce platform for a client.
| What | How |
|---|---|
| Backend | Node.js, Express, REST APIs, JWT auth, FastAPI · learning Spring Boot |
| Frontend | React, Tailwind CSS, Vite |
| Databases | MongoDB, MySQL |
| AI / LLMs | Multi-agent pipelines, Gemini and Groq APIs, local models with Ollama and Whisper |
| Infra | GitHub Actions CI/CD, Render, Vercel |
| Languages | Java, C++, Python, JavaScript |
- Shipping: projects that are deployed and used, not just demoed.
- LLMs with guardrails: spend caps, trace logs and evals in CI, so AI features stay predictable.
- Local-first AI: tools that run on the user's own machine when the data is personal.
- Fundamentals: 270+ LeetCode problems and counting.
- 💼 SDE roles (Class of 2027)
- 🤖 Data / AI engineering roles
- 🌱 Open-source collaboration
- 🤝 Hackathon teams
