Hi, I'm Rushabh
For 10+ years I've turned machine learning into working products across healthcare, insurance, telecom and banking. Today I lead GenAI engineering for risk and controls at Citi, building RAG and multi-agent systems with responsible-AI guardrails.
My career runs like an agent loop: plan, call a tool, learn, repeat. Pick a route and I'll run it.
Worked and studied with
Roadmap
Education or experience: choose one and watch my agent run it. Every stop is one turn of a real agent loop: plan, think, call a tool, check the result, remember it.
agent.route("tell my story")
Two branches, one agent
Run a whole branch, or jump straight to any stop on it. Both branches merge where I am today.
in: what is already in memory · out: the industry this step lands in
Output
The final answer: GenAI that is useful, measurable and safe enough for a bank, built with every toolkit the run picked up along the way.
May 2025 – now · New York
Citigroup · Controls Technology
I lead a globally distributed team of GenAI, UI and backend engineers. We design RAG, agentic and multi-agent systems for risk and controls, own the group's GenAI roadmap, and build responsible-AI guardrails into every deployment.
Evals
Every agent needs evals. Mine are LinkedIn recommendations from managers, teammates and professors. Tap anyone in the orbit to read their feedback.
Research
From Bluetooth payments in Pune to protein language models.
Say hello
The loop keeps running. If you are working on GenAI in a regulated space, I would love to compare notes.