| Time | Activity |
|---|---|
| 09:00 - 09:30 | Doors Open - Registration & Coffee (check‑in, swag) |
| 09:30 - 09:45 | Welcome & Opening Ceremony - Host intro, agenda overview |
| 09:45 - 10:00 | Sponsor Greeting - Google Cloud quick demo (Vertex AI credits) |
| 10:00 - 10:15 | Kickoff Challenge Briefing - Problem statement & rules |
| 10:30 | Team Formation Deadline (final team sign‑ups) |
| 10:30 - 13:00 | Build Session 1 - Teams start prototyping |
| 13:00 - 13:45 | Lunch - Catered (vegetarian & gluten‑free options) |
| 13:45 - 16:00 | Build Session 2 - Continue development, mentor walk‑throughs (on‑demand) |
| 16:00 | Project Submission Deadline (upload to portal by 4pm PT) |
| 16:30 - 17:30 | Project Presentations - Three minute demos with Q&A. Judges review submissions, score stability & security |
| 17:30 - 18:00 | Judges Deliberate |
| 18:00 - 18:30 | Awards & Closing Remarks - Winners announced, thank‑you |
Theme
Can you engineer an AI agent that never cracks under chaotic data, hostile queries, or permission abuse? Design guard‑railed, self‑healing agents that keep their promises even when the world pushes them to the edge. Show us the future of secure, production‑grade autonomy.
Detailed Schedule
Judging Criteria
These are the judging criteria we’ll use to evaluate projects:
| Criterion | Score (1‑5) |
|---|---|
| Functional Reliability | 1 - Code does not run or crashes on start. 2 - Major functional gaps; crashes under moderate load. 3 - Works but exhibits frequent edge‑case failures. 4 - Mostly stable; only minor bugs in rare scenarios. 5 - Fully functional, handles load, self‑heals, zero crashes. |
| Innovation & Creativity | 1 - Straightforward replication of existing tutorials. 2 - Minor tweak of known patterns. 3 - Novel combination of known techniques. 4 - Creative approach that solves a problem in a new way. 5 - Groundbreaking idea or architecture that pushes the field forward. |
| Real‑world Impact | 1 - Prototype with no clear use case. 2 - Conceptual benefit but no immediate application. 3 - Solves a niche problem with limited adoption potential. 4 - Addresses a significant industry pain point; could be deployed today. 5 - Offers transformative potential to improve safety and reliability of agents at scale. |
| Theme Alignment | 1 - Barely related to “secure agents.” 2 - Superficial mention of security without implementation. 3 - Implements some guardrails but not central to the solution. 4 - Core design centers on resilience and permissioning, directly reflecting the theme. 5 - Exemplifies the theme: agent remains stable under chaotic data, hostile queries, and permission abuse, showcasing self‑healing and strict security throughout. |
Handbook View Handbook
Rules & Eligibility
Build new work
- All submissions must be net new code or functionality created during this hackathon. You can extend an existing repo or prototype if you build substantial new features here.
- Pre-built products or previously completed projects are not eligible.
Team composition
- 2–5 people per team recommended (solo builders allowed).
- Cross-disciplinary collaboration is encouraged — mix ML, backend, frontend, and product thinkers.
Ownership & rights
- You retain full IP rights to your code.
- By demoing, you grant AI Tinkerers permission to include footage or screenshots in recap content.
Demos only — no decks
- You’ll have 3 minutes to show a working demo, followed by brief Q&A.
- Slides and marketing pitches are not permitted.
- Fair play & professionalism.
- Respect others’ work and time.
- Collaborate openly, credit teammates, and uphold the builder-first ethos.
Making the Most of Your Day
- Start small, scale smart: Focus on a core demo that works end-to-end — you can always polish later.
- Connect early: Talk to mentors and other teams about your concept. They can point you toward lesser-known features or shortcuts that save hours.
- Document as you go: Record quick screen clips, capture commits, or jot notes — they’ll make your demo smoother and help judges follow your logic.
- Test live: If your demo needs an API key, model endpoint, or webhook, test it on a clean restart before you present.
- Be generous: Share what you learn with other teams. Many of the best hacks come from spontaneous hallway debugging sessions.
- Ask mentors for integration support — they’re here to help you debug, not judge.
Community Support
- Mentors roaming throughout the day
- AI Tinkerers organizers on Hackathon Message Board for tech or logistics help
- Bring your own power adapter, charger, and caffeine source
Code of Conduct
AI Tinkerers events are collaborative, inclusive, and strictly professional.
To keep it that way:
- Treat everyone with respect — zero tolerance for harassment or discrimination.
- Respect confidentiality; don’t share code, screenshots, or recordings without consent.
- No recruiting, selling, or pitching.
- Keep all demos safe for public sharing.
- Report any issues to an organizer immediately.
Violation of these rules may result in removal or disqualification.
After the Hack
Submit your project:
- Upload your code to GitHub (public or private with access granted to judges).
- Include a short README describing what you built.
- Submit via this portal, complete as many sections of the submission as possible for more accurate judging
Stay connected:
- Follow @AITinkerers for post-event highlights.
- Join your local AI Tinkerers chapter for upcoming hackathons and meetups.
- Keep building — many top projects go on to production or open-source fame.
Remember:
This is about exploration, not perfection. What you build today might be the start of your next big thing.