Selected Work
From a voice-controlled MCP agent to a self-serve print kiosk SaaS — systems designed end-to-end, from architecture decisions to shipped code.
Core Project
A voice and text AI agent that controls any registered software via natural language, using a custom REST-to-MCP bridge, multi-software tool chaining, and a destructive-action confirmation gate. Also the foundation of my MCA final year project, "Voice-Driven Software Automation Using MCP-Integrated AI Agent."
Core Project
A fully on-premise SaaS product for copy centers across India: customers scan a QR code, upload a PDF, pay via UPI, and the shop's own PC silently prints the file — no cloud server, no customer data leaving the shop's network.
More Work
Voice-enabled AI interview practice platform — FastAPI/React with screening, Socratic-method, and case-discussion modes, powered by Groq Llama and pgvector.
GitHub ↗A voice and text medical chatbot that uses RAG to search uploaded medical documents — React/Vite frontend, FastAPI backend, Groq Whisper STT, and pgvector on Neon PostgreSQL.
A Backstage.io-inspired internal developer portal — Angular 20, Firebase, NgRx, D3.js dependency graphs, and Gemini AI, deployed live.
Interested?