Tanmay Sah, PhD

Tanmay Sah, PhD works on applied AI evaluation, model validation, and the reliability of autonomous agents. His research examines safety–success tradeoffs in tool-using language-model agents, and he presented his doctoral work on the subject at the inaugural ACM Conference on AI and Agentic Systems. He also co-developed TanML, an open-source toolkit for validating tabular machine-learning models.

Career

Cosigned byLoading...

CosignedSign in to view

Based in: Minneapolis

ARB Trading Group

Nov 2013 — Jul 2018

Quantitative Developer( Software Engineer)

Harrisburg University of Science and Technology

PhD

National Institute of Technology Warangal

Bachelor's studies

North Carolina State University

Master's studies

Products & media

ai.engineer

Your Coding Agent Can't Undo Its Own Mistakes

Tanmay Sah and Dolly Sah explain UndoBench and EvoUndo, research on whether coding agents can recover from failures and reverse their own changes.

Media appearance

ai.engineer

1,368 AI Failures and What They Teach About Evals

Sah and Vishal Srivastava discuss what real-world AI failures reveal about evaluating deployed systems, including retrieval, tools, human review, permissions, and recovery.

Media appearance

joss.theoj.org

TanML: Automated Model Validation Toolkit for Tabular Machine Learning Models

A Journal of Open Source Software paper describing TanML, an automated validation toolkit for tabular machine-learning models, coauthored by Sah.

Writing

digitalcommons.harrisburgu.edu

Evaluation in Applied AI: Predictive Modeling, Agent Safety, and Model Validation

Sah's doctoral dissertation covers research on social-media virality, safety evaluation for tool-using agents, and the TanML model-validation toolkit.

Writing

Profile name

Company name

Company name

Profile name

Company name

Company name

Profile name

Company name

Company name

Profile name

Company name

Company name

Profile name

Company name

Company name

Tanmay Sah, PhD

ARB Trading Group

Harrisburg University of Science and Technology

National Institute of Technology Warangal

North Carolina State University

Products & media

Your Coding Agent Can't Undo Its Own Mistakes

1,368 AI Failures and What They Teach About Evals

TanML: Automated Model Validation Toolkit for Tabular Machine Learning Models

Evaluation in Applied AI: Predictive Modeling, Agent Safety, and Model Validation

Curated Lists

List Inclusions