Dr. Tanmay Sah, PhD
PhD in Data Science • 14+ Years Industry Experience

Building reliable AI systems.

I am an AI researcher and quantitative modeler specializing in machine learning, model validation, large language models, AI agents, verification, and neural routing architectures.

GitHub: @tradertanmay Google Scholar LinkedIn

Selected Research & Key Projects

Core research initiatives, open-source toolkits, and publications in AI agents, verification, and neural routing.

Conference Presentations & Invited Talks

Research presentations, papers, and keynotes delivered at major artificial intelligence and data science conferences.

ACM CAIS 2026 Main Research Track

ACM Conference on AI and Agentic Systems

Paper Presentation: "The Verifier Tax: Safety-Success Tradeoffs in Tool Using LLM Agents"

Presented empirical research examining safety-success tradeoffs in tool-using LLM agent execution, introducing formal verifier tax metrics to mitigate agentic execution risks.

Authors: T. Sah, V. Srivastava, D. Sah, K. Jordan Watch Presentation ↗
AI Engineer World's Fair Workshop Session

AI Engineer World's Fair

Workshop: "LLM Inference at Scale"

Conducted technical workshop on scaling LLM inference, memory optimization (KV Caching, PagedAttention), quantization, and high-throughput model serving infrastructure.

Topic: Scalable LLM Inference & Optimization View Official Schedule ↗
SciPy Conference Conference Mentor

SciPy Conference

Conference Attendee & Mentor

Attended the SciPy Scientific Computing in Python conference and served as a mentor for a conference attendee in scientific computing and Python data science workflows.

Topic: Scientific Computing & Python Mentorship
ODSC AI West Training Session

ODSC AI West (Open Data Science)

Training Session: "Serving LLMs Efficiently: A Hands-On Journey from Single-GPU Basics to Distributed Inference"

Will be conducting hands-on training session covering efficient LLM serving, single-GPU optimization strategies, memory management (KV Caching, PagedAttention), quantization, and multi-GPU distributed inference pipelines.

Topic: Distributed LLM Inference & High-Throughput Serving
NC State University Quant Workshop

North Carolina State University

Invited Speaker: "TanML Model Validation & Development"

Presented TanML automated model validation and development framework to graduate researchers and practitioners in the Master in Financial Mathematics Program Quant Workshop.

Topic: Automated Model Validation (TanML) View Quant Workshop ↗