An opportunity for developers to build the next wave of on-device AI applications is presented through the Qualcomm x Meta ExecuTorch Hackathon. This event centers on creating innovative, responsive, and privacy-aware AI experiences by leveraging the power of PyTorch, ExecuTorch, and Snapdragon-powered mobile hardware. Participants will have the chance to work directly with industry experts, test their creations on real devices, and compete for prizes that support further development.
Event Details
• Format: On-site only, with no online or hybrid participation.
• Location: San Francisco
• Dates: June 27–28, 2026
• Capacity: Limited to 150 on-site participants.
• Team Requirement: Teams must consist of 3–5 members.
• Application Deadline: June 15, 2026
• Selection Process: Applications undergo a preliminary screening process.
Hackathon Focus
• Real Mobile Hardware: Build and test applications on Snapdragon-powered Samsung Galaxy S25 Ultra devices provided during the event.
• Low Latency: Create fast, responsive AI experiences that run locally on the device instead of relying on the cloud.
• Privacy-Aware Development: Explore the possibilities that emerge when AI processing occurs directly on the user's device.
• Expert Support: Receive hands-on guidance from Qualcomm and Meta experts to facilitate progress and resolve technical challenges.
Challenge Overview
• Main Challenge: Create a responsive, user-facing AI application that executes locally on mobile hardware using ExecuTorch on Snapdragon. Projects should demonstrate the benefits of on-device AI, such as lower latency, enhanced privacy, offline functionality, or energy efficiency.
• Build for Real Mobile Usage: Applications should be practical for phone use, focusing on real-time user interaction.
• Optimize for Edge Performance: Focus on speed, responsiveness, NPU utilization, and energy efficiency on Snapdragon-powered hardware.
• Meaningful Local Processing: Highlight how running AI locally improves privacy, reduces cloud dependency, or creates a better user experience.
• Suggested Project Directions: Computer vision, multimodal assistants, generative AI on mobile, privacy-first AI, and real-time edge intelligence.
Technology & Resources
• Core Technology Partners: Qualcomm provides the mobile AI platform and developer tooling, while Meta supports the PyTorch to ExecuTorch workflow.
• Development Workflow: Build with PyTorch, deploy with ExecuTorch, and run on Snapdragon mobile hardware.
• Testing Hardware: Prototypes will be built and validated on Snapdragon-powered Samsung Galaxy S25 Ultra devices.
• Developer Resources: Participants will have access to Qualcomm Developer Resources, Qualcomm AI Hub, DevRel Edge AI Sample Apps, third-party tools, and resources from previous hackathons.
Event Structure
• Check-in & Device Access: Teams check in on-site and receive access to devices and tools.
• Kick-off & Masterclass: The event starts with sponsor introductions and a technical overview of the core technologies.
• Hands-on Building: Teams spend the event building, optimizing, and testing their on-device AI applications.
• Mentor Support: Industry experts will be available on-site to provide guidance and support.
• Live App Demos: Each team will demo their application, highlighting its technology, performance, and user experience.
• Judging & Awards: Projects are evaluated by judges, with an additional Team’s Choice Award selected by participant vote.
• Additional Perks: The event includes meals, participant swag, networking opportunities, and a closing social hour.
Prizes & Awards
• Top Prize (Selected by Judges): Each member of the winning team receives a Meta Quest 3 512GB headset. The team also gains access to Qualcomm DevRel support, App Store publishing support, and opportunities for features in blogs and live streams.
• Team’s Choice Award (Popular Vote): Each member of the winning team receives Ray-Ban Meta AI Glasses, along with Qualcomm DevRel support, App Store publishing support, and blog and live stream opportunities.
Submission Requirements
• Basic Information: Project title, short and long descriptions, and technology tags.
• Presentation Materials: A cover image, video presentation, and slide presentation.
• Code & Hosting: A public GitHub repository, demo application platform, and application URL.
• Repository & Code Rules: The application must be in a public GitHub repository, contain no closed-source existing code, and be open for public consumption.
• README Requirements: The README must include an application description, team member names and emails, setup and usage instructions, and an open-source license.
• Application Readiness: The application must primarily run on the edge, be installable on its intended platforms, and function as described.
Judging Criteria
• Technical Implementation (40%): Evaluation of NPU utilization, latency, performance, and energy efficiency.
• Application Use Case & Innovation (25%): Assessment of problem-solving, creativity, uniqueness, and user experience.
• Local Processing & Privacy (15%): Focus on the extent of on-device execution and privacy or security benefits.
• Deployment & Accessibility (10%): Judged on the ease of installation and use.
• Presentation & Documentation (10%): Based on the clarity of the presentation and the quality of code and documentation.
Participation & Application
• Eligible Participants: The event is designed for developers, engineers, AI builders, and product-minded teams.
• Team Formation: Participants are expected to build as a team of 3-5 members. Assistance is available for individuals looking for teammates.
• How to Participate: Interested teams must submit the application form by June 15, 2026, for preliminary screening. Pre-approved participants will receive further instructions to complete their registration.
This hackathon provides a focused, hands-on environment for builders to push the boundaries of mobile AI. By bringing together top talent with expert guidance and cutting-edge hardware, the event aims to accelerate the development of practical, high-performance applications that run where users are: on-device.