Amitesh Gupta

I’m an AI-focused backend engineer, currently pursuing a B.Tech in Computer Science and Engineering at IIIT Kota. I like turning complex prompts and real-time events into production-grade systems that are fast, reliable, and easy to extend.

In my AI / Backend Engineer Intern role at Gyanama, I architected a real-time multi-language voice AI agent with Twilio + an LLM, hitting sub-1-second per-turn latency via a custom 8kHz to 24kHz audio pipeline. I also built a 3-tier attendance escalation pipeline with post-call NLP summary extraction for automated ERP write-back, and an assignment generation service using FastAPI + LanceDB + an LLM for RAG-powered PDF quiz creation.

To make the system dependable at scale, I implemented an event-driven Redis Streams sync layer for 6+ event types across distributed Python and Node.js services, backed by dual-MongoDB for strict data isolation. I resolved 8 critical production incidents across WebSocket security, Twilio signatures, and LLM response resilience—improving reliability and uptime.

My projects reflect the same focus: CodeSync AI uses Monaco + Y.js CRDTs for collaborative code editing with sub-100ms sync latency, while also shipping a Cursor-style RAG assistant with SSE-streamed responses and inline citations. Algorhythm pairs LangGraph orchestration with deterministic LLM parsing (regex + JSON fallbacks) to produce interactive HTML visualizations with strong parse success and runtime stability.