Uploaded October 2024 | Updated September 2026, 2 weeks ago

---

Coinbase's shift from Sagemaker to Ray marks a watershed moment in the company's ML infrastructure implementation. This session unveils how Coinbase dramatically accelerated its ML training processes, slashing iteration times from hours to mere seconds and scaling to terabyte-level datasets.

ML Platform Engineer Wenyue Liu details Coinbase's journey, highlighting the challenges with Sagemaker and the strategic migration to Ray. Learn how this transition not only boosted training speeds and scale but also delivered over 20% cost reduction in instance usage.

For ML practitioners and engineers seeking to elevate their training infrastructure, this session provides a roadmap to achieving unprecedented efficiency and scale in ML operations.

--

Interested in more?
- Watch the full Day 1 Keynote: youtu.be/jwZHJthQvXo
- Watch the full Day 2 Keynote youtu.be/Lury2ad6KG8

--

🔗 Connect with us:
- Subscribe to our YouTube channel: youtube.com/@anyscale
- Twitter: https://x.com/anyscalecompute
- LinkedIn: linkedin.com/company/joinanyscale
- Website: anyscale.com