Finalist Verification

Before you award the prize,
verify what was actually built.

Independent technical verification for hackathon finalists.

I check submitted claims against code, demos, deployments, and required technology, then prepare cited evidence for your judges.

Human judges still decide.

See what gets verified

The scope

What gets verified

For each finalist, I investigate the claims that matter to the award decision.

  1. Does the project actually run?
  2. Does the repository support the submitted claims?
  3. Is the required technology actually being used?
  4. Does the demo match the implementation?
  5. Can important architecture or deployment claims be verified?
  6. What remains unclear or needs judge follow-up?

The deliverable

What you receive

A concise evidence packet for each finalist.

Example evidence packet

Humor Genome Studio

Illustrative findings, not a completed verification.

Claim Evidence Result Source
Uses Gemma in the core generation flow Repository + runtime inspection Verified Relevant files and observed runtime behavior
Core generation flow runs from the submitted repository Setup + launch + sample generation Verified Setup instructions and captured runtime output
All analysis functions shown in the demo are implemented Demo compared with available code paths Partially verified Demo segments and corresponding functions; some paths remain untested
Deployed version matches the submitted repository Deployment compared with the submitted revision Not independently verified No build or commit reference provided for the deployment

Not independently verified means the available evidence was insufficient to confirm a claim. It does not mean the claim is false.

Questions for judges

  • Where exactly is Gemma invoked?
  • Which analysis functions were demonstrated in the submitted build?
  • Does the deployed version match the submitted repository?

The boundary

Verification, not judging.

sound.fan does not choose winners or generate final scores.

It investigates submitted evidence and reports findings as Verified, Partially verified, Not independently verified, or Inconsistent.

AI investigates evidence.
Human judges decide.

When to use it

After finalists are selected and before winners are announced.

Send the finalist submissions, judging rubric, and relevant artifacts. I independently verify the technical claims that matter to the award decision and return cited findings for your judges.

  1. Finalists selected
  2. Evidence reviewed
  3. Judges review
  4. Winners decided

Work with me

Founding Finalist
Verification Pilot

  • Up to 10 finalists
  • Technical claims checked against submitted artifacts
  • Runtime verification where practical
  • Cited evidence packet for every finalist
  • Open questions for judges
  • Human judges retain full control

$1,000

Founding pilot pricing while the service is being validated with early events.

The experience

Built from both sides of the judging table.

I organized and reviewed submissions for Build with Gemma: Humor Genome NYC, informed by firsthand experience across 60+ hackathons and competitions.

FAQ

Common questions

Before you announce the winners
When should I use finalist verification?
After finalists are selected and before winners are announced.
What do you need from us?
The rubric, finalist submissions, and access to relevant artifacts such as repos, demos, or deployments.
Can you work with private repos or materials?
Yes, if access and handling expectations are agreed in advance.
What happens if something can’t be verified?
I report what is supported, what remains unclear, and what judges may want to follow up on. Human judges still decide.
Do you verify events you participate in?
No. I do not provide pre-award verification for a hackathon I am participating in.

Finalist Verification

Request finalist verification

Ready for verification, planning ahead, or have a question? Share what you know. The founding pilot is $1,000 for up to 10 finalists.