2022 Sky Camp

Wednesday, October 19, 2022 – Banatao Auditorium, Sutardja Dai Hall, UC Berkeley Campus

Sky Camp is where you can get exposure to specific Sky Computing research projects through talks and demos, as well as hands-on experience with systems and technologies for emerging applications. This includes a new multi-region and multi-cloud data transfer service; a platform for collaborating with confidential data, privacy-preserving data analytics, and learning; a compiler for automated high-performance distributed deep learning on-top of JAX and soon PyTorch; and a new platform for running model training scripts across cloud providers.

Bookmark this page for :

Recordings

Watch the Sky Camp recordings here!

Agenda

Times listed are in PDT

  • 9:00 AM – 10:00 AM: Breakfast
  • 10:00 AM – 10:20 AM: Opening Talk: Overview of Sky Camp
  • 10:20 AM – 12:00 PM: Skyplane Tutorial
    • Skyplane is a tool for blazingly fast bulk data transfers in the cloud. Skyplane manages parallelism, data partitioning, and network paths to optimize data transfers, and can also spin up VM instances to increase transfer throughput.
    • Skyplane: Slides
    • Skyplane Storage: Slides
  • 12:00 PM – 1:00 PM: Lunch
  • 1:00 PM – 2:30 PM: SkyPilot Tutorial
    • In this tutorial, we’ll introduce SkyPilot, a tool to seamlessly run jobs across multiple clouds while minimizing your costs. We will use SkyPilot to secure spot GPU instances to train a ML model, inject spot instance preemptions and use SkyPilot’s failure recovery mechanisms to transparently recover and resume training.
    • Slides
  • 2:30 PM – 2:45 PM: Break
  • 2:45 PM – 4:00 PM: Alpa Demo
    • Scaling neural networks to hundreds of billions of parameters has enabled dramatic breakthroughs such as GPT-3, but training and serving these large-scale neural networks require complicated distributed system techniques. Alpa aims to automate large-scale distributed training and serving with just a few lines of code.
    • Slides
  • 4:00 PM – 4:15 PM: Break
  • 4:15 PM – 5:30 PM: MC2 Tutorial
    • MC2 is a platform for running secure analytics and machine learning on encrypted data. With MC2, users can outsource their confidential data workloads to the cloud, while ensuring that the data is never exposed unencrypted to the cloud provider. MC2 also enables secure collaboration — multiple data owners can use the platform to jointly analyze their collective data, without revealing their individual data to each other.
    • Tutorial
    • Slides
  • 5:30 PM – 8:00 PM: Reception at the Wozniak Lounge, Soda Hall

Contact

For questions email skycamp [at] berkeley.edu

FAQ

Is there parking?

We strongly encourage public transportation or ride share. However we can provide parking passes for the nearby garage, Upper Hearst Parking Structure. The garages tend to fill up by around 10:00 AM so please arrive early. If you will need a parking pass, please email us.

How can remote attendees participate?

Remote attendees can view the livestream and participate in the Sky Camp Slack conversations on the day of event. Tutorials will be supported for onsite attendees only.

Covid Precautions

Covid tests will be available for attendees to use at their discretion. Masks are strongly recommended throughout the event.

How do I register?

Registration is by invite only. If you think you should have received an invitation please email us.

Prerequisites / Technical Requirements

  • “Modern” browser (Firefox, Chrome, Safari etc.)
  • Experience programming in Python in notebook environment
  • Basic understanding of AI/ML concepts (e.g., training, validation, linear models)

Lodging

Some local hotel options are listed on our contact page: https://sky.cs.berkeley.edu/contact/