Student guide // judging
How your project is judged
Saturday, November 21, 2026 · ITKAN Innovation Hub · 8:30 AM – 7:00 PM
Your team is scored out of 100: 35 points for the idea and 65 points for the build and live demo. The judges score with exactly this rubric, so use it to choose and shape your idea now.
Scoring at a glance
Five criteria. 100 points.
| Part | Criterion | Weight | The question judges ask |
|---|---|---|---|
| Idea | Problem & impact | 20% | Is this a real problem for a real user, and would solving it make a difference? |
| Idea | Originality | 15% | Is this a fresh idea, or a fresh take on a known one? |
| Build & demo | Working demo | 25% | Does it run live, in front of us, doing what the team says it does? |
| Build & demo | Quality of AI use | 25% | Is AI doing real work, in the product or in how it was built? |
| Build & demo | Presentation | 15% | Did the team explain it clearly, on time, and handle questions well? |
A great idea that doesn't run can't win. A working demo of a vague idea will struggle too.
The rubric
What a weak, solid and excellent score looks like
Judges give each criterion a whole score from 1 to 5. A 2 or 4 sits between the descriptions below.
| Criterion | 1 Weak | 3 Solid | 5 Excellent |
|---|---|---|---|
| Problem & impact20% | Problem is vague or invented; unclear who would use it | Real problem and a clear user, but thin on how much it would help them | A specific user the team knows or has talked to, evidence the problem is real, and a believable path to people actually using it |
| Originality15% | A copy of a tutorial project or a common app (another to-do list, another generic chatbot) | A known idea with one meaningful twist | Judges haven't seen it before, or it solves a known problem in a surprising way |
| Working demo25% | Doesn't run, or only slides and screenshots are shown | Core flow runs live, with rough edges or one part faked and admitted | The full core flow runs live on a judge's own input, and handles bad input and edge cases |
| Quality of AI use25% | AI is bolted on and adds nothing, or the team can't explain what the AI did | AI does real work, inside the product or in how the team built it | AI does work that would be hard without it; the team can explain which model, why, roughly what it costs, where it fails, how they check its output, and how they protect user data |
| Presentation15% | Over time, hard to follow, one person answers everything | Clear story, on time, answers most questions | Problem, demo and result land in 4 minutes; direct answers; every member can explain their part and what they learned |
A product with no AI inside it can still earn a 5 on Quality of AI use if you can show you used AI tools exceptionally well to build it.
Before you submit your idea
Six questions
Check your idea against these six questions. Each one maps to points on the rubric.
- 01
Who exactly is it for? Name the user. Better still, talk to one before the event.
- 02
How do you know the problem is real? A story, a number, or something you've seen yourself.
- 03
What's new about it? If a free app already does it, what's your twist?
- 04
Can the core flow run in 7.5 hours? Cut it down to the one thing that must work live.
- 05
Where does AI do real work? Not a chat box added at the end.
- 06
What could go wrong with the AI? Wrong answers, private data, bias. Have a plan for it.
Building and submitting
Your project
Details on the AI tools you'll use, and how you'll submit your finished project, will be sent to all participants before the event.
Either way, be ready to explain two things in your demo: how you used AI to build it, and anything that's faked or unfinished.
The demo and Q&A
4 minutes, then 3 of questions
Each team gets 4 minutes to demo and 3 minutes of questions. Demos start at 5:15 PM.
- It has to run live. Recorded videos and slides of screenshots don't count as working.
- Judges may ask to try it with their own input.
- The timer is strict. At 4 minutes the demo stops and Q&A begins. Practice first.
- Show what's real. Admitting something is faked costs a little. Hiding it and being found out costs a lot.
- Everyone should be able to answer. Judges may ask any team member.
- Demos are recorded so the panel can review them.
Rules
Break one, and it’s more than points
Breaking one of these can get a team disqualified, not just marked down.
- 01
Teams of 2 to 5, formed before the event. Changes on the day need the organizer's approval.
- 02
Start fresh at 9:20 AM. Planning, research, sketches and design beforehand are encouraged. Building the project before 9:20 AM is not.
- 03
Hard stop at 5:00 PM. Nothing changed after 5:00 PM counts.
- 04
Open-source libraries, APIs and public datasets are allowed. Name them in your demo.
- 05
Your work is your own. Teams own what they build. ITKAN may use names, photos and demo recordings for publicity.
Results and prizes
Out of 100, averaged
Your final score is the average of the judges' scores, out of 100; ties go to the higher Working demo score. Results are announced within 48 hours, and judges' decisions are final.
Prizes per team
ITKAN AI Hackathon · Student guide to judging · Version of October 14, 2026
← Back to the AI Hackathon