Kwasi Ankomah
AI architect focused on inference and agentic enterprise AI
Palo Alto, California
Summary
Expert in high-performance AI inference and agentic AI architectures — focuses on making large models run efficiently in production, enabling deep multi-step agents and fast model switching for enterprise use cases. wandb+2
Builds practical evaluation and LLMOps tooling for enterprises — helped develop an evaluation jumpstart kit and integrates logging and evaluation with Weights & Biases to benchmark speed, cost, and accuracy across models. github+1
Experience applying AI in regulated and enterprise domains — background includes roles in financial services, government, and consulting, bringing data-led deep learning approaches to complex, regulated problems. stacresearch+2
Active public speaker and educator — frequently speaks on inference, AI infrastructure, and responsible AI at industry events, webinars, and podcasts. brighttalk+2
Work
Education
Projects
Writing
Same Model, Three Platforms: What Function Calling Benchmarks Reveal
December 1, 2025Examines how infrastructure differences across inference providers affect function calling and structured output accuracy for identical models; presents benchmarks and implications for production reliability.
From Insight to Action with SambaNova Agents
October 1, 2025Discusses agentic AI workflows on SambaNova's platform and how fast, efficient inference enables agentic applications and evaluation at scale.