Hi, I'm Shrinish Vhanbatte 👋

Backend + AI systems engineer — Member of Technical Staff at TestMuAI(Formerly LambdaTest), building high-throughput microservices and AI-powered testing infrastructure. Dual Degree (B.Tech + M.Tech), IIT Kharagpur.

I like problems where distributed systems and LLMs meet: agents that call real tools, eval pipelines that keep them honest, and the queues, caches, and databases underneath.

🔭 What I'm building

• AgentCI — CI/CD & eval platform for AI agents. Eval suites fan out across a model × parameter matrix, get scored by a two-stage LLM-as-judge, and produce baseline diffs, flakiness scores, and a pass/warn/fail CI gate. FastAPI control plane (REST + WebSocket + MCP server), Kafka workers, Postgres + ClickHouse.
• Fivo — voice-first personal-finance agent. Tool-calling agent that manages accounts, transactions, and budgets from speech, backed by a RAG pipeline on pgvector with Voyage AI embeddings. Async FastAPI, SQLAlchemy 2, Redis, Celery.

💼 What I do at work

• Engineered a high-performance Visual Engine (C++/Python, AWS Lambda, Docker) powering smart image comparisons — boosted product revenue by 20%
• Built Go microservices with Kafka-driven event flows: scheduler service, CRUD APIs, multi-baseline screenshot comparison with Redis-coordinated goroutine fan-out
• Added network-log capture to KaneAI real-device sessions via a MITM proxy in the HyperExecute VM flow, powering AI test authoring

🛠 Toolbox

Go Python C++ TypeScript SQL
FastAPI Kafka Redis MySQL PostgreSQL/pgvector ClickHouse AWS Docker Kubernetes
LLM agents & tool calling RAG LLM-as-judge evals MCP LangChain Datadog

🏆 Elsewhere

• LeetCode 2140 (top 1.46%) · Codeforces 1529 · CodeChef 1812 — handle: bhatte
• Global rank 865 in Google Kick Start Round H 2022 · rank 1531 in Meta Hacker Cup Round 2
• 🥇 Gold, Inter-IIT Tech Meet 10.0 — large-scale system design (Mercari)
• Mentored 10+ students via KOSS · contributor to Rocket.Chat

📫 Reach me

LinkedIn · [email protected]