Dr. Sharon Zhou
Corporate VP of AI
AMD
Presentation Title:
Building trustworthy domain-specific AI agents
Presentation Summary:
LLMs are powerful but their non-deterministic nature makes steering them toward accuracy a challenge.
In this session, we’ll provide a blueprint for building trustworthy domain-specific AI agents that breaks down each of the following components:
- Generating high-quality synthetic training datasets
- Leveraging different LLM adaption techniques based on your domain and use case
- Creating agentic evaluation pipelines
- Iterating to higher levels of accuracy and complexity
About | Sharon Zhou
Dr. Sharon Zhou is VP of AI at AMD and was the Founder and CEO of Lamini, an AI startup that won the VentureBeat Gen AI Startup Award and was recognized as a Forbes Cloud 100 Rising Star. As a former Stanford faculty member, she led a 50+ person Generative AI research group and published award-winning papers in the field.
Sharon teaches top AI courses on Coursera, including Fine-tuning LLMs and Diffusion Models, reaching millions of developers and executives. She earned her PhD in AI from Stanford, advised by Dr. Andrew Ng. Before her PhD, she worked as an AI Product Manager at Google. She holds a BA in Computer Science and Classics from Harvard.
Sharon has also served as an AI advisor in Washington, D.C. and was named to MIT Technology Review’s 35 Under 35 list.