Join Qualcomm and Google, with sponsorship from Samsung, for a hands-on hackathon designed to help developers build, deploy, and optimize on-device AI experiences using Google LiteRT on Snapdragon® mobile platforms.

This event is focused on making it easy to go from model to mobile application—leveraging production‑ready tools, prebuilt assets, and reference implementations to accelerate development and deployment on Snapdragon® mobile devices.

About the challenge

Participants will choose from one of the following tracks, each designed to minimize setup friction and maximize time spent building meaningful AI experiences. Each team of 3-5 will receive one Samsung Galaxy S25 Ultra to use for the duration of the hackathon.

 
Track 1: LiteRT-LM - LLM-Based Consumer Use Journeys

Build on-device LLM experiences using latest Gemma models with a focus on real consumer use cases.

What’s provided:

  • Ready-to-use the latest Gemma models, EmbeddingGemma, FastVLM  available via LiteRT Hugging Face or equivalent sources
  • A reference sample app based on Gallery App - see  Assets for more details

This track is ideal for developers interested in integrating LLM-powered features into Android applications with on-device inference.

 

Track 2: LiteRT - Classical Models — Vision & Audio

Build efficient, on-device AI applications using classical ML models for vision or audio workloads.

What’s provided:

  • Models that do not require quantization or conversion via Qualcomm® AI Hub (optionally LiteRT Hugging Face models)
  • NPU-friendly models available via Qualcomm® AI Hub or the LiteRT Hugging Face Model Zoo
  • A sample application available through the LiteRT repository or shared source code

This track is ideal for developers looking to deploy performant vision or audio pipelines on Qualcomm Hexagon NPUs with minimal overhead.

Why Participate

  • Learn how to deploy on-device AI efficiently using LiteRT and Qualcomm Hexagon NPU
  • Get hands-on experience with production-ready models, tools, and sample apps
  • Build alongside engineers and experts from Qualcomm Technologies and Google
  • Walk away with working code, practical knowledge, and confidence to bring AI features into real products

Whether you’re experimenting with on-device LLMs or optimizing classical ML workloads, this hackathon is designed to help you ship faster—and smarter—on Snapdragon.

Snapdragon and Qualcomm branded products are products of Qualcomm Technologies, Inc.

Requirements

Project Requirements

  • What to Create: Entrants must create a working software application that leverages Google LiteRT to deliver an on-device AI experience optimized to run on a Qualcomm Snapdragon-powered mobile device, entered into one of the following tracks (each a “Project”):

    1. LLM-Based Consumer Use Journeys — Build on-device LLM experiences using Gemma 4 models with a focus on real consumer use cases.

    2. Classical Models — Vision & Audio — Build efficient, on-device AI applications using classical ML models for vision or audio workloads.

  • Functionality: 

    1. The Project must be capable of being successfully installed and running consistently on the platform for which it is intended and must function as depicted in the video and/or expressed in the text description.

    2. The application must be developed and/or commercially ready to the extent that it may be deployed on app store or other open source platform for users to download.

  • Platforms: A submitted Project must run on the platform for which it is intended and which is specified in the Submission Requirements. 

  • New Projects Only: Projects must be newly created by the Entrant after the start of the Hackathon Submission Period.

  • Third Party Integrations: If a Project integrates any third-party SDK, APIs and/or data, Entrant must be authorized to use them in accordance with any terms and conditions or licensing requirements of the tool.

Submission Requirements 

Submissions to the Hackathon must meet the following requirements:

  • Include a Project built with the required developer tools and meets the above Project Requirements.

  • Provide a URL to your code repository in Github for judging and testing. The Github repository must contain all necessary source code, assets, and instructions required for the project to be functional. The repository must be public and open source by including an open source license file (please refer to https://choosealicense.com for help choosing). This license should be detectable and visible at the top of the repository page (in the About section). The repository should also include:

    • A README file with the following information:

    • An application description;

    • Names and emails of all Eligible Individuals on the Team;

    • Setup instructions from scratch, including dependencies if applicable; and

    • Run and usage instructions

  • Include a text description that should explain the features and functionality of your Project.

  • Optional: The following components are highly recommended but not mandatory:

    • Tests and testing instructions to verify the app setup;

    • Notes section containing additional information not covered in the application description;

    • References used while developing the application; and

    • Well-commented code.

Hackathon Sponsors

Prizes

2 non-cash prizes
Judge's Top Choice
1 winner

Winner of the judge's vote
-Each member of the team will receive a Nintendo Switch 2 console

Participant's Top Choice
1 winner

Winner of the participant's popular choice award
-Each member of the team will receive a Nintendo Switch 2 console

Devpost Achievements

Submitting to this hackathon could earn you:

Judges

Derrick Johnson

Derrick Johnson
Qualcomm Technologies, Inc.

Aaishika Bhattacharya
Qualcomm Technologies, Inc.

Felix Baum

Felix Baum
Qualcomm Technologies, Inc.

Uttam Tripathi

Uttam Tripathi
Qualcomm Technologies, Inc.

Chintan Parikh

Chintan Parikh
Google

Derek Bekebrede

Derek Bekebrede
Google

Matthias Grundmann

Matthias Grundmann
Google

Cormac Brick

Cormac Brick
Google

Judging Criteria

  • i. Technical Implementation
    Evaluation based on NPU utilization, latency and performance, and energy efficiency.
  • ii. Application Use-Case and Innovation
    Evaluation through the lens of problem solving, creativity and uniqueness, and user experience.
  • iv. Deployment and Accessibility
    Evaluation based on ease of installation and use.
  • v. Presentation and Documentation
    Evaluation based on the clarity of explanation during the presentation, and code quality and documentation.

Questions? Email the hackathon manager

Tell your friends

Hackathon sponsors

This site is protected by reCAPTCHA and the Google Privacy Policy and Terms of Service apply.