A Snowflake hackathon at Data in the D — Friday, October 16, 2026 — Detroit
Every data team has work that eats hours and never gets done: the analysis nobody has time for, the manual grind, the request that lands in your inbox every week. This hackathon is about using AI to get rid of it.
You get five hours of open build time and a pre-provisioned Snowflake account with datasets loaded and AI features turned on. Click a link and start building. There's no environment setup, so you don't lose the first hour. You keep the account for 30 days after the event, so if you start something good, you can keep going.
You'll leave with something you built yourself and a clear picture of how AI agents fit into data work.
Who it's for
Data analysts, data engineers, and data-curious practitioners from any industry, at any skill level. New to Snowflake? A 2-hour guided workshop runs alongside the hack so you can ramp up first.
Three tracks, judged separately
- Hack the Question (analysts): The analysis nobody has time to do. Use AI to understand, clean, and analyze messy data, then deliver a visual report a decision-maker can act on.
- Hack the Grind (engineers): The work you shouldn't be doing by hand. Build agents, automations, and tools that create, change, or send something: pipeline monitors, triage tools, reusable team skills.
- Hack the Experience (builders): What your end user actually touches. Build a conversational interface, an app, or a self-serve tool for someone who doesn't write SQL.
Schedule — Friday, October 16 (ET)
| Time | |
|---|---|
| 11:00 AM | Registration |
| 12:00 PM | Hacking begins · Guided workshop & demos run in parallel |
| 5:00 PM | Event concludes · Winners announced |
| 6:30–8:30 PM | Speaker & Sponsor Welcome Reception |
Prizes
One winner per track. Prize details coming soon.
Requirements
What to build
A working project for one of the three tracks, built on Snowflake. Every project must:
- Include a visual component. A Streamlit app, HTML report, dashboard, or something else people can interact with.
- Use several Snowflake features that work together. See the Resources tab for recommended features by track.
- Use AI to do real work. It can't just be decoration.
Pick the right track
- Hack the Question: Read-only. The output is a report, briefing, or finding that a person reads and acts on. If there's a button that changes data, enter Hack the Grind.
- Hack the Grind: Something gets created, changed, or sent. If nothing does, enter Hack the Question.
- Hack the Experience: The main user is someone other than you and your team. If that user could have built it themselves, it's the wrong track.
What to submit on Devpost (by 5 PM ET)
- Project name and the track you're entering
- Short description: the problem, who it's for, and what you built
- The Snowflake features and AI capabilities you used
- Screenshots or a short demo video (under 3 minutes)
- An honest "How Baked Is It" self-assessment (scale below) that says what's real and what's faked. Judges reward accuracy over confidence.
- Optional: link to your code repository
Data you can use
The standard event dataset, your own data, Snowflake Marketplace datasets, or synthetic data you generate on the spot.
Judging criteria
| Criterion | Hack the Question | Hack the Grind | Hack the Experience |
|---|---|---|---|
| Insight Quality | 25% | — | — |
| Communication | 15% | — | — |
| Time Reclaimed | — | 15% | — |
| Reusability | — | 15% | — |
| Experience Quality | — | — | 35% |
| Relevance | — | — | 25% |
| How Baked Is It | 20% | 20% | 20% |
| Platform Craft | 20% | 30% | 10% |
| Meaningful AI Use | 20% | 20% | 10% |
The "How Baked Is It" scale
- Storyboard: The screens exist, but the logic doesn't.
- Happy path only: It only works if you follow the demo script.
- Working prototype: It runs end to end on real data, with rough edges.
- Usable by someone else: It handles bad input and needs no tour.
- Handoff-ready: It could go to a real user on Monday.
Most strong five-hour projects score a 3. A 4 is genuinely impressive.
Prizes
Raspberry Pi 5 8GB Kit 64GB Edition
1 winner of the Raspberry Pi
Beats Studio Pro Wireless Headphones
Beats Studio Pro Wireless Headphones
Bose SoundLink Plus Portable Bluetooth Speaker
Bose SoundLink Plus Portable Bluetooth Speaker
Devpost Achievements
Submitting to this hackathon could earn you:
Judges
Luke Hartzell
Gilberto Hernandez
James Cha-Earley
Judging Criteria
-
How Baked Is It (all tracks)
Demo-ware, prototype, or ready to hand to a user Monday? 1 Storyboard, 2 Happy path only, 3 Working prototype, 4 Usable by someone else, 5 Handoff-ready. Most strong projects land at 3. Reward honest self-scores over confidence. 20% all tracks. -
Platform Craft (all tracks)
Did the team combine several Snowflake features deliberately, or just tick boxes on a checklist? Weight: 20% Question, 30% Grind, 10% Experience. -
Meaningful AI Use (all tracks)
Did AI do real work here, or was it decoration? Weight: 20% Question, 20% Grind, 10% Experience. -
Insight Quality (Hack the Question only)
Did they find something true and non-obvious, or restate what everyone already knew? The strongest work brings in outside context or surfaces something counterintuitive. Weight: 25%. -
Communication (Hack the Question only)
Could a decision-maker read this and act on it without someone to translate? Weight: 15%. -
Time Reclaimed (Hack the Grind only)
Does it take real hours off a real, named workflow, not a hypothetical one? Weight: 15%. -
Reusability (Hack the Grind only)
Could someone outside the team pick this up and use it? Weight: 15%. -
Experience Quality (Hack the Experience only)
Would a real end user succeed with this, and enjoy using it? Weight: 35%. -
Relevance (Hack the Experience only)
Does it adapt to who the user is, or serve everyone the same thing? Weight: 25%.
Questions? Email the hackathon manager
Invite others to compete
This site is protected by reCAPTCHA and the Google Privacy Policy and Terms of Service apply.