Tanmay Sah, PhD

Tanmay Sah, PhD Tanmay Sah works at the intersection of machine learning, model risk, AI evaluation, and agentic AI systems. His work focuses on building, evaluating, and validating reliable AI systems across predictive modeling, credit risk, and AI agents, with research centered on applied AI evaluation, tool-using LLM agents, and safety–success tradeoffs in real-world AI systems. He is the author of “The Verifier Tax: Horizon-Dependent Safety–Success Tradeoffs in Tool-Using LLM Agents,” accepted at ACM CAIS 2026. His broader research interests include LLM evaluation, multi-agent systems, verifier architectures, agent safety, AI reliability, and scalable inference. He has also published research on machine learning for social media virality and created TanML, an open-source machine learning toolkit designed to simplify ML model validation. He is especially interested in the next generation of trustworthy AI systems: agents that can reason, use tools, remain auditable, and operate safely under real-world constraints.

Career