Call for Participants: GA4GH “Next Steps for Federated Learning” Hackathon
Reggan Thomas
Dear colleagues,
We are planning a hackathon for the “Next steps for federated learning and GA4GH” session on April 15th at Connect! This will be a great opportunity to explore some of the toughest questions associated with federated computation–via novel concepts, pull requests, and prototypes.
To structure the hackathon, we have identified several challenge areas that reflect key associated with federated machine learning (FedML) in federated environments. This is not an exhaustive list, and we welcome ideas for new challenge areas or topics.
Proposed Challenge Areas
1. Access to Data and Compute Resources
Lead: TBD
Possible hackathon topics:
Authentication and authorization passthrough across federated services (e.g., task-specific tokens, visas)
Write-back mechanisms for FedML updates or results, potentially aligned with work in the Cloud Work Stream around DRS write-back
Prototyping a plugin interface for workflow engines such as Funnel, including passing Task Execution Service (TES) packets via HTTP headers for authentication/authorization
Exploring token handling logic that sits in front of compute runtimes
Investigating how compute resources can be securely accessed across federated services and how trust relationships are established
2. Billing and Cost Accounting
Lead: TBD
Possible hackathon topics:
Approaches for tracking and distributing compute (WES/TES) costs across a (heterogeneous) federated ecosystem
Handling data egress costs in distributed environments, building on recent updates to DRS metadata and regionalization
3. Telemetry and Service Information
Lead: TBD
Possible hackathon topics:
Mechanisms to discover available compute resources across endpoints
Exposing infrastructure capabilities (e.g., GPUs, runtime environments) through service-info APIs to support smarter workload placement
4. Policy Implementation
Lead: TBD
Possible hackathon topics:
Coordination with relevant workstreams to explore implementation of policy frameworks
Developing guidelines, templates, or reference agreements that could help organizations establish federated collaborations more easily
Over the coming weeks we will prepare collaboration spaces (Git repositories, shared folders, and communication channels) so that participants can work efficiently during the hackathon sessions.
We need your help! We are currently looking for:
Challenge area leads to lead discussions and hackathon teams
Suggestions for additional challenge areas or hackathon topics
RSVPs – Please indicate your interest in participating in the hackathon!
Please feel free to express your interest here by March 20, or share your thoughts in the Federated ML Slack channel (#federated-ml)
Many thanks, and we look forward to your participation.
Regards,
On behalf of Andrew, Kyle and Venkat
Reggan Thomas
Dear colleagues,
A reminder that the Federated Learning Hackathon for the “Next steps for federated learning and GA4GH” session will take place on April 15th at Connect! This is a unique opportunity to tackle some of the toughest questions in federated computation through novel concepts, prototypes, and pull requests.
To structure the hackathon, we have identified several challenge areas reflecting key aspects of federated machine learning (FedML) in federated environments. This is not exhaustive, and we welcome ideas for additional challenge areas or topics.
Proposed Challenge Areas:
Access to Data and Compute Resources – Lead: TBD
Possible topics include authentication/authorization passthrough, write-back mechanisms for FedML updates, plugin interfaces for workflow engines (e.g., Funnel), token handling logic, and secure access across federated services.
Billing and Cost Accounting – Lead: TBD
Possible topics include tracking and distributing compute costs across federated ecosystems and handling data egress costs.
Telemetry and Service Information – Lead: TBD
Possible topics include discovering available compute resources across endpoints and exposing infrastructure capabilities through service-info APIs for smarter workload placement.
Policy Implementation – Lead: TBD
Possible topics include exploring policy frameworks, developing guidelines, templates, or reference agreements to support federated collaborations.Real-World Implementation – Lead: TBD
Connect to APPFL TES network from a local laptop or compute network.
We will prepare collaboration spaces (Git repositories, shared folders, and communication channels) over the coming weeks to help participants work efficiently during the hackathon.
We need your help! We are currently looking for:
- Challenge area leads to lead discussions and hackathon teams
- Suggestions for additional challenge areas or hackathon topics
- RSVPs – Please indicate your interest in participating in the hackathon!
Please express your interest by tomorrow, March 20 here or share your thoughts in the Federated ML Slack channel (#federated-ml)
Many thanks, and we look forward to your participation!
Please note there is no obligation to respond to emails from me outside of your typical working hours.