SIH 2025 National Winner · Building LLMOps Infrastructure · Prompt Ops · Eval Pipelines · Agentic Systems · Mumbai
Bachelor of Engineering
University of Mumbai
Join Tanish and 10k+ other professionals on
Tal is the app where bosses meet engineers & builders directly, without middlemen
Download Tal BossSIH 2025 National Winner · Building LLMOps Infrastructure · Prompt Ops · Eval Pipelines · Agentic Systems · Mumbai
I build AI systems that are skeptical of themselves. Most of my projects start from a real frustration with how LLM-based systems fail silently — RAG that hallucinates confidently, agents that loop without learning, pipelines with no quality gate. I build the layer that catches those failures before they matter. Recent work: a Corrective RAG system that routes queries through a 3-verdict confidence engine before ever touching the LLM (K-State); a multi-agent research pipeline with an automated review-retry loop that won't ship output scoring below 7/10 (Research Paper Analyser); and Chronicle, a prompt version control platform with a Pareto frontier eval engine that tells you exactly what accuracy you're buying at what cost. I won Smart India Hackathon 2025 (1st nationwide) for SimAstra — a decentralized Hunter-Killer drone swarm system built for Bharat Electronics Limited. No central controller, communication-denied environments, 100+ simulated combat scenarios. My stack centers on Python, LangChain, LangGraph, CrewAI, FastAPI, and Groq. I care a lot about LLMOps — versioning, eval, observability, and the gap between a demo that works and a system that holds up. Currently finishing my BE in Computer Science at Theem College of Engineering (May 2026) and actively looking for AI/ML or backend engineering roles where the problems are hard and the bar for "it works" is high. If you're building something real with LLMs or agents — let's talk.
Mumbai
Bachelor of Engineering, University of Mumbai