Sr. TPM, ServiceNow Previously Frinks.AI IIT Guwahati Hyderabad
Evaluation systems for AI that has to be trusted inside a company.
I ship scoring, review and the operational loops that decide whether an AI judgment is allowed to change how engineering actually works. Not demos. Production gates, explainable scores and the failure modes you only see at org scale.
Selected work
A 0–5 AI score with explainable subscores for whether a story is actually ready to build. Replicated as Epic Quality. Now a leadership signal, not a dashboard widget.
Direction and evaluation framework for LLM code review. The product is the contract: 5% false positives or the tool does not get to sit in the review path.
Flagship visual-inspection AI. MVP in 1.5 months, 50% faster time-to-market, launch through compliance. Inspection only works if false rejects and escaped defects are treated as product, not model trivia.