Autopilot for Algorithms: Optimizing HPC Code and Mixed-Precision Performance with AlphaEvolve
October 13, 2026
9:00 AM - 10:00 AM Pacific Time
Online
Embedded Files
About the Session
For advanced computing professionals, manual performance tuning is one of the most frustrating bottlenecks in scientific discovery and software development. AlphaEvolve shifts code tuning from an art form to an automated, AI-driven science.
This session demonstrates how to leverage AlphaEvolve on Google Cloud to automatically refactor serial, parallel, and hardware-accelerated codebases. We will explore how to set up automated evolutionary loops, optimize for highly efficient mixed-precision math, and review how leading enterprises like Substrate have successfully slashed compute costs and skyrocketed code execution speeds. Attendees will leave with the step-by-step developer guidelines needed to use AlphaEvolve for their own HPC code optimization.
Speakers
Jay Boisseau
Advanced Computing Strategist, Google Cloud
Jay Boisseau is an advanced computing strategist at Google, where he manages a strategic initiative to make Google Cloud the best platform for scientific/technical computing and leads the Google Cloud Advanced Computing Community. He is an experienced leader in HPC and advanced computing technologies, with 30 years of experience developing and leading programs in academia and industry. His previous experience includes founding and leading the Texas Advanced Computing Center to global prominence, and providing strategic planning and special projects in HPC at Dell Technologies. Jay also leads the Austin Forum on Technology & Society, founded the Austin AI Alliance, and is the lead founder and owner of Remedy bar in Austin (where he conducts lots of tech meetings and events so it aligns). He earned his doctorate from UT Austin in astronomy, which led him into supercomputing and technologies.
Fadi Hamad
Software Engineer, High Performance Computing, Google Cloud
Google Sites
Report abuse