Topics I am happy to speak about. I have not given a public talk yet, and I would like to. If you are organizing a KubeCon, KCD, or meetup and want a session on Kubernetes scheduling, DRA, or running ML infrastructure at scale, reach out.

Topics I speak on

• Dynamic Resource Allocation (DRA): writing a driver, and migrating from device plugins.
• Multi-node inference and distributed training: LeaderWorkerSet, JobSet, and gang scheduling.
• Running multi-tenant, multi-cloud Kubernetes platforms at scale.
• The Kubernetes scheduler: how it works and how to extend it.

Past talks

None yet. When I give one, it will show up here with the event, the date, and a link to the slides or video.

Related writing

The deep dives behind these topics:

• Writing a DRA driver
• LeaderWorkerSet for multi-node inference
• JobSet for distributed training
• Gang scheduling in 1.35 and 1.36
• Kube-scheduler internals and writing a plugin