Ray on Kubernetes: Powering Quant Research at Scale | Ray Summit 2024

preview_player
Показать описание
Quantitative trading firms are constantly pushing the boundaries of computational power and efficiency. This session reveals how Ray on Kubernetes is revolutionizing quantitative research infrastructure, enabling massive scalability and unifying diverse workloads.

Dizhou Wu shares insights from building a cutting-edge platform that supports up to 100,000 concurrent CPU cores and seamlessly extends to GPUs. Wu explores the integration of Ray into quantitative research workflows, bridging the gap between large offline research jobs and interactive intraday tasks. Wu also delves into the challenges of running Ray on Kubernetes with spot instances, presenting a custom Ray controller solution that builds reliability on inherently unstable infrastructure. Beyond the technical aspects, attendees will gain valuable perspectives on introducing Ray within an organization, from securing buy-in to effective user education strategies.

--

Interested in more?

--

🔗 Connect with us:
Рекомендации по теме