When it was brought up on our internal OCP team call recently that we should streamline our event terminology, I jumped at the opportunity. I’m nothing if not a fan of efficiency and consolidation. Our event team took some time over the last few days to review our Summit lingo and made some co (...)
Hackathon
San Jose, California
- Home
- Registration
- Travel
- Schedule Overview
- Full Schedule
- Schedule Highlights
- Sponsors
- Speakers
- Experience Center
- Hackathon
- Co-Located Events
- Symposium
2023 Edge-Native AI Hackathon
Hosted by OCP, LF Edge and ETSI
Hosts
![]() |
![]() |
![]() |
Sponsors
![]() |
![]() |
Edge computing, including cloud provider regional data centers, communication service provider wired and wireless points of presence and on premises enterprise equipment continues to emerge with use cases that require workloads distributed from centralized data centers out to edge locations. The objective is to provide low latency response times, minimize bandwidth consumed transmitting data, and respecting data sovereignty, while dynamically optimizing placement of workloads. To make matters even more interesting AI workloads will be placed at the edge as well as within centralized data centers.
For this reason, the OCP, LF Edge and ETSI have chosen edge native AI/HW/SW Co-design as the theme for an upcoming Hackathon. Three application vertical categories, as shown in the figure below, for the Hackathon that are especially relevant for Edge-Native design, use of AI, and HW/SW co-design:
- AI and Automation: How to utilize edge-native and AI into data analytics and automation?
- Sustainability: How is edge-native and AI useful to enhance sustainability in applications, services, and networks?
- 5G Edge: How to deploy and utilize AI in 5G edge infrastructure to enable edge-native distributed applications?
The edge with Edge-Native AI applications presents special challenges:
- The Edge offers local, low-latency resources to architects and developers to create a new class of “Edge-Native” applications and services. These Edge-Native solutions are specifically designed to take full advantage of the distributed nature of Edge Networks.
- Increasingly, the Edge is offering advanced resources beyond compute and storage, including access to GPUs, AI models & tools, real-time context data, and analytics.
- AI HW/SW Co-design is fundamentally required to match the capabilities of form platforms to services. This feedback loop is intended to provide application developers optimal flexibility to continue to innovate and maximize system performance.
- In this Hackathon, teams are challenged to design and prototype an Edge-Native solution, focusing on the use of AI and considering overall application performance optimization. Key considerations include:
- The Edge is not homogenous, offering a range of resources and hardware capabilities from on-premises edge nodes → to the mobile network edge → to city or regional edge data centers. And let’s not forget about the Cloud and its capabilities.
- A fundamental design challenge of any Edge-native solution is to determine what functionality is best run at each location distributed from cloud to edge in the system at any given-time and if functionality may need to transition from one location to another based on application requirements, device mobility, end-user quality of experience, and availability of edge resources.
- Your solution may include any combination of client applications, edge applications and services, and cloud components.
- For example, if a solution needs generative AI (such as a chatbot), how to optimize performance of large language models (LLM) between a centralized cloud data center vs, an edge region, or vs. an on-prem edge node with their varying HW capabilities and network latencies.
- How are we defining performance for the Hackathon? We are not. Performance depends on the context and needs of the solution. Hence, we are asking teams to define performance in their context and share their insights on how AI/HW/SW co-design influences it.
HACKATHON RESOURCES (offered to competing teams)
- The OCP offers two Cloud environments:
- Qarnot (https://qarnot.com/en) (https://tasq.qarnot.com/
login/) - A PaaS/SaaS platform: Hackathon teams can easily deploy HPC/batch computation tasks with docker containers.
- A Bare Metal platform: Hackathon teams will be provided with full SSH access to OCP recognized servers (even IPMI / L2 if necessary)
- Flax Computing (https://www.flaxcomputing.
com/sustainable-server-labs) - To support AI, Hackathon teams can be provided with OCP recognized GPU servers with 8x K90 GPUs each, as well as NVMe storage nodes.
- For 5G Edge, Hackathon teams can be provided with OCP recognized servers, with Wifi option
- Qarnot (https://qarnot.com/en) (https://tasq.qarnot.com/
- The LF Edge offers
- Akraino blueprints:
- https://wiki.akraino.org/display/AK/Smart+Cities
- https://wiki.akraino.org/display/AK/KubeEdge+Edge+Service+Blueprint
- https://wiki.akraino.org/display/AK/The+AI+Edge+Blueprint+Family
- https://wiki.akraino.org/display/AK/CFN+%28Computing+Force+Network%29+Ubiquitous+Computing+Force+Scheduling
- https://wiki.akraino.org/display/AK/CPS+Robot+Blueprint+family
- https://wiki.akraino.org/display/AK/5G+MEC+System+Blueprint+Family
- Akraino Platform Security Architecture -- Provides core knowledge about platform security, requirements, and its importance for Edge solutions. It may give some ideas for improving solutions that will be developed during the Edge Hackathon.
- LF Edge Catalog - Provides (potential hierarchy of) recipes for deployment of templates
- LF Edge Community Lab - Hackathon teams can reserve compute and networking resources in the lab using their Linux Foundation ID and the lab dashboard. Lab resources include x86_64 and arm64 systems, with installed GPUs and configurable networking fabrics.
- Akraino blueprints:
- ETSI MEC offers a set of open and standardized Edge Service APIs to enable development and deployment of edge-native applications at scale. For this Hackathon, the following MEC services are applicable. OpenAPI representations are available and development ready on forge.etsi.org
- MEC011 - Edge Platform Application Enablement (Mp1): MEC App and Service Support and Management (e.g., MEC application registration, MEC service advertisement and discovery)
- MEC012 - Radio Network Information Service (RNIS): 4G / 5G radio network contextual info
- MEC013 - Location Service: network and geo-location information about terminals and infrastructure, etc.
- MEC015 - Bandwidth Management Service & Multi-access Traffic Steering Service
- MEC016 - Device Application Interface: service that enables terminal devices to discovery MEC applications and create application context
- MEC021 - Application Mobility Service (AMS): relocation of user context between MEC application instances across deployed MEC platform
- MEC028 - WLAN Access Information Service (WAIS): WiFi network contextual info
- MEC030 – V2X Information Service (VIS): predicted quality of service for vehicle routes.
- Refer to these ETSI White Papers for helpful information on developing using ETSI MEC Service APIs:
- Hackathon teams will also be provided with access to the ETSI MEC Sandbox. The MEC Sandbox is an online environment where developers can access and interact with live ETSI MEC Service APIs available in an emulated edge network set in Monaco, which includes 4G, 5G, & WiFi networks configurations, single & dual MEC platform deployments, etc.
The Hackathon In-Person Event: 2023 OCP Global Summit
At the culmination of the remote phase of the hackathon, three finalists will be selected for the final in-person competition, to be held at the OCP Global Summit in San Jose on October 17-19. The three finalist hackathon teams are to showcase their solutions on the Expo Hall Floor of the OCP Global Summit, where the finalists have the opportunity to connect with global industry leaders participating in the event. In a Final Pitch-Off, the three finalist teams will present their solutions in front of a live audience and a judging panel representing ETSI, LF Edge, OCP, and others. Each of the top three teams will receive a $2,000 travel stipend.
The winners are to receive:
First Place: $3,000
Second Place: $2,000
Third Place: $1,000
DEADLINES
Submission Period OPEN: July 1
Submission Period CLOSED: July 21
Notifications of acceptance: July 24
Hackathon Start - Remote Competition: July 24 - September 15
Team Final Submission - September 15
Three finalist teams notified week of September 25
Hackathon in-person event at OCP Global Summit: October 17-19
Team size: 5 members (maximum)
Price: The submission is free of charge, as well as the Hackathon participation for the selected teams, and includes free admission to the entire 3-day 2023 OCP Global Summit.
Copyright: We do not require any assignment of copyrights for this Hackathon.
Open-Source: We are also not requiring teams to open source their developed code. However, we encourage teams to do so. Sharing code in open source will help us in judging submissions.
SPONSORSHIP OPPORTUNITIES
We are looking for sponsors to participate in the Hackathon!
- Prize Sponsor - $10,000 (includes the cash prizes)
- First place award - $3,000
- Second place award - $2,000
- Third place award - $1,000
- Travel Sponsor - covers a small stipend for team travel - $10,000
- Equipment Sponsors - Opportunities to provide materials that the developers will use for the hackathon, such as:
- x86/ARM/RISC-V platforms suited for development with AI, including potentially accelerator/GPU options, high memory options, availability to access high storage capacity
- Edge specific platforms such as Open Edge Servers and such
- Any enabling open APIs, tools, data sets, and so on
Sponsors receive the following:
- Logo exposure on event website, email blasts, collateral and social media posts
- Onsite signage
- 5-minute speaking spot at the award announcement (no slides - open to first 3 sponsors)
- Participate as judges for the Hackathon
Contact dirkv@opencompute.org if you are interested in sponsoring this exciting event.