Accelerating Frontier AI: The Compute Crisis is a Reinforcement Learning Problem
In this event, you'll learn how RL is being used to design more efficient infrastructure, run more efficient training, and squeeze more capability out of existing compute so you can tackle the escalating energy and cost problems of frontier AI at every layer of the stack.
π Event Overview
Join us at the AWS Builder Loft in San Francisco for a session on reinforcement learning, compute optimization, and the economics of frontier AI. In this event, you'll learn how RL is being used to design more efficient infrastructure, run more efficient training, and squeeze more capability out of existing compute so you can tackle the escalating energy and cost problems of frontier AI at every layer of the stack.
ποΈ Event Details
- π Location: AWS Builder Loft, San Francisco
- π Date: β οΈ [Please confirm β not shown on the Luma page]
- β° Time: 5:00 PM β 8:00 PM PT
- π Register Here: Accelerating Frontier AI β DeepAI San FranciscoΒ
π What You'll Learn - By the end, you will be able to:
- Understand how RL agents design chip floorplans, generate and tune GPU kernels, control cooling for gigawatt-scale AI factories, and optimize the training process itself
- Recognize where these approaches can be pushed further, and how continual learning may reshape efficient foundation model training
π€ Who Should Attend - This event is designed for:
- Researchers and engineers in reinforcement learning
- Engineers working on foundation model training
- Builders working with AI compute infrastructure β and anyone whose GPU bill keeps them up at night
π Event Format - This session will include:
- A brief introduction to key concepts
- Guest talks (incl. Rick Gentry, InstaDeep; Recurvia training-acceleration demo) and hands-on learning
- A panel discussion, plus time for questions, discussion, and a social hour with the RL and frontier AI community
π Prerequisites - To get the most out of this session:
- You should be familiar with reinforcement learning, foundation model training, or AI compute infrastructure
- Make sure to bring a valid government-issued photo ID (mandatory, no exceptions) and pre-check-in via the AWS Builder Loft event spaceΒ
π Why This Event Matters
As the energy and cost of training frontier AI continues to grow β with training runs measured in hundreds of millions of dollars and data centers in gigawatts continues to grow, understanding how to use RL to get more from the compute we already have is becoming increasingly important. This session is designed to give you practical experience you can apply to your own work.
π¬ Final Notes! - Spots are limited. Register now to secure your spot!
Here is the registration link again: Accelerating Frontier AI β DeepAI San FranciscoΒ
π Ready to see what else you unlocked? Head to builder.aws.comΒ to complete your profile, check out upcoming Loft events, and start exploring.
π See you at the Loft! We're at 525 Market St, 2nd Floor, San Francisco. Quick heads up: you'll need a government-issued photo ID and you have to be 18+ to enter. Also, we don't have parking for bikes or scooters, so find street parking nearby (they're not allowed in the building! sorry!).
Enjoyed reading this content? Let the author know!
Your likes, comments, shares, and saves help creators reach more builders.
Loading recommendations
Loading article