VELOby excentrix

proof of work · not promises

Anyone can generate a project. 
Can you defend it? 

AI made polished output worthless as a signal. VELO runs an adaptive interrogation of your own code — and turns whether you can defend it into a verifiable credential you can share.

For developers · hiring teams · colleges · Built in Bangalore

to a verifiable proof-of-work credential
5 min
to a verifiable proof-of-work credential
adaptive questions on your own code
6–10
adaptive questions on your own code
ways to fake a verdict you can’t defend
0
ways to fake a verdict you can’t defend

why velo exists

In 2026, a perfect repo proves nothing.

AI can write the project, polish the README, and pad the resume. Green squares, clean commits, a flawless portfolio — all of it is now generatable in an afternoon. The signal employers relied on for a decade is dead.

What can’t be faked is defending your own work under questions that adapt to your answers. That’s the one thing VELO measures — and the one thing that still means something.

how it works

Five minutes from repo to credential.

No gameable test. An adaptive interrogation that goes wherever your answers do.

  1. 01

    Point us at your work

    Paste a GitHub repo and a one-line claim about what you built. No setup, no installs.

  2. 02

    Liveness check

    VELO checks the repository is real, yours, and matches the claim — before a human ever looks.

  3. 03

    Adaptive interrogation

    An AI examiner asks 6–10 questions about your own code, probing deeper wherever an answer is vague. You can’t pre-write your way through it.

  4. 04

    A verdict you can share

    You get a defensibility score and a public, verifiable proof-of-work credential — the thing that actually belongs on your resume.

what an interrogation looks like

You can’t pre-write your way through it.

Every question adapts to your last answer. Vague replies get probed harder. This is a real exchange about a side project’s rate limiter.

live verification · session #4127

VELO

Your README says the rate limiter is “token bucket”. Walk me through what happens when two requests arrive in the same millisecond on different servers.

Candidate

Each server has its own bucket in memory, so they’d both be allowed — the limit is per-instance, not global.

VELO

So the limit you advertise as “100/min” isn’t actually enforced globally. What would you change to make it global, and what’s the cost?

Candidate

Move the counter to Redis with an atomic INCR + expiry. Cost is a network hop per request and a Redis dependency on the hot path.

VELO

Good — that’s a real answer, not a memorised one. Moving on.

✓ Defended · contributes to a verified proof-of-work credential

one engine · three sides

pricing

One engine. Pay for what you verify.

You never pay for unlimited anything that runs an interrogation — pricing maps to real verifications, so the margin holds however much anyone uses it.

Developer

Freethen ₹399/mo

Prove what you built.

  • First verification free
  • Shareable proof-of-work credential
  • Public verification page
  • Re-verify as you ship more
Verify a project

Hiring teams

Creditspay per verification

Know who actually built it.

  • Send candidates a verify link
  • Defensibility verdict per candidate
  • Compare against your own read
  • No seat minimums
Start free pilot

Colleges

Per seatper student / semester

Send graduates out with proof.

  • Verify whole cohorts
  • Placement-ready credentials
  • Cohort dashboard + reporting
  • Onboarding support
Talk to us

start free

Verify your first project.

Drop your email and the repo you’re proudest of. We’ll open your interrogation slot and send you a shareable, verifiable proof-of-work credential.

Free for your first verification · No card required