OranBook 30 min
← selected work

Product, 0 → 1

Skipy, the v1 of Traast

Skipy was our MVP: a hiring manager describes a role once, and meets only people who already passed an interview. I wrote the PRD, built it myself, put it in production with real users, then handed our CTO a product that already worked. He rebuilt it as Traast, which went to production with real customers.

Watch the demo

A job post on Monday becomes a first interview on Friday

New tab

I built this demo as one self-contained page: no video, no backend. Companies, contacts and numbers are fictional; the mechanics are real.

Best viewed on a laptop: Open the full demo

soloPRD → working v1
13screens
→ AWShandover, Lovable to AWS
wiredpayments
3analytics stacks wired

Context

A startup without a recruiter posts one job and gets buried in hundreds of resumes that only the hiring manager can judge. Skipy flips that: the hiring manager describes the role once, and meets only a handful of people who already passed an interview. Every recruiting tool starts from the pile; this one starts from the role and skips the pile.

What I built

  1. Specified the workflow, states, business rules and acceptance tests; wrote the PRD and built 13 screens in Base44, then Lovable
  2. An AI role builder: a job post in, a structured brief out, with the missing 90-day goals drafted
  3. An ops console for our talent team: AI fit analysis against the brief, then pre-interviews
  4. One click invites the chosen candidates; they accept on a no-login page and book through Calendly into Google Calendar
  5. Defined integration authentication, field mapping and failure behavior for the Supabase, SendGrid, Slack and payment connections
  6. Three analytics stacks wired in (Mixpanel, PostHog, Google Analytics) to watch real users, not guess
  7. Validated the workflow with users, then handed our CTO a working product as the implementation reference for Traast

How it runs

Skipy ran in production with real users. Its successor, Traast, ran on AWS with real customers. Alongside the product work, I trained co-founders and employees to use AI workflows in daily operations.

What it taught me

A working prototype is a better spec than any document. The CTO didn't need a 40-page PRD; he needed a product that already had users clicking through it and analytics showing where they got stuck.

Stack

Want the walkthrough?

Company product. Demo above is public; the codebase is not.

Book 30 minoranov.con@gmail.com
Next · GTM engineeringThe Lead Machine →