Kumar Sanu

I’m an AI Backend Engineer focused on building high-performance LLM applications—especially LLMs, RAG pipelines, and multi-agent systems—paired with robust backend architecture. I’ve worked end-to-end from orchestration (LangChain/LangGraph) to production-ready services that improve cost and responsiveness.

In my AI Engineer role, I built Mirai, an AI-powered learning platform used by students and teachers. I developed FastAPI + MongoDB backends with Gemini File Search and LiveKit, and deployed LLM pipelines for notes, summaries, Q&A, flashcards, cheetsheets, and podcasts at scale.

Previously, I scaled multi-tenant ERP and websites for thousands of schools, supporting 99.9% uptime and optimizing APIs with Redis caching, SQL tuning, and rate limiting to cut latency by 60%. On the AI side, I delivered an AI Hinglish learning assistant with 85% engagement, using RAG with Redis vectors to reduce redundant LLM calls by 40% and improve latency by 25%.