Rules

  1. Code must be written after the hackathon starts, 7/15 8am.
  2. All coding must stop at 7/16 2:30pm.
  3. Projects will only be judged on working code. They must be more than a pitch, presentation, or clickable prototype.
  4. Projects may be built on existing projects, but! Judging will only based on new features or ported open-source generative models. Forking or building on projects must also be mentioned in the demo to the judges.

Judging

  1. Teams will be assigned to a room (posted on Discord): either 407, 454, or 462.
  2. Teams will give a 2-minute presentation and/or live demo of their project to judges.
  3. Judges will pick finalists to present to everyone (posted on Discord).
  4. Out of the finalists, judges will pick 1st, 2nd, and 3rd prize winners + side-bounty winners.
Qualcomm Bounty

The Qualcomm judge will pick submissions that use the Qualcomm Innovators Development Kit - QIDK. It is highly recommended to attend the Qualcomm workshop to learn more about how to use QIDK.

  1. To get started, there are 5 sample apps that walk through using QDIK https://github.com/quic/qidk/tree/master/Solutions
  2. One possible submission is porting new foundational model over to QDIK and adding the the sample app repository
Replit Bounty

The Replit team will judge submissions that use Replit after the hackathon on July 17. We’ll announce the winners in Discord and on social media.

To be considered for this bounty:

  1. Indicate the project is built on Replit in the submission form on Devpost

  2. Have a repl link included in the "try it out" section in your Devpost project.

  3. Optionally, include a short demo video of your project.

Prizes

See the hacker handbook.

Judging Criteria

You will be graded on each of these categories:

  • Completion: Do you have a working prototype?
  • Creativity: Has project been done before? How creative is it in solving the problem?
  • Design: Is UI attractive and easy to use? Or if command line, is it well documented?
  • Usage of AI technology: How hard was the problem to solve? How did AI make project possible?