1
0 Comments

How I Solo Built Complete Interview Prep (From Idea to Users)

I built Complete Interview Prep solo.
No co-founder.
No agency.
No VC money.

Just a very specific problem I had personally—and a lot of stubborn execution.

This post is a breakdown of why I built it, how I built it & what worked. If you’re an indie hacker building alone, this is for you.


The Problem I Experienced First-Hand

I had already built another product in the same space: Game Assessment Prep, focused on HireVue game-based assessments.

What I kept seeing from users was consistent:

  • They could practice games
  • They could understand logic and scoring
  • But video interviews were still killing them

Not because they were unprepared on content—but because:

  • They didn’t know how they looked on camera
  • They didn’t know how answers were actually evaluated
  • They had no realistic way to practice under pressure

Most “interview prep” products are:

  • Static PDFs
  • Generic questions
  • Human coaches that don’t scale and cost too much

I wanted something different:

A product that simulates the interview, evaluates you like a recruiter, and gives actionable feedback—automatically.

That’s how Complete Interview Prep started.


Why I Built It Solo

Short answer: speed and control.

Long answer:

  • I already had domain expertise (interview assessments)
  • I already knew the user pain extremely well
  • I didn’t want alignment discussions before validating demand
  • I wanted full control over UX, pricing, and iteration speed

Solo building meant:

  • Faster decisions
  • Fewer compromises
  • Brutal accountability

If it failed, it was on me.
If it worked, I’d know exactly why.


The Core Product (MVP Scope)

I forced myself to keep the scope tight.

Complete Interview Prep does only one thing well:

It lets users practice realistic interviews and get structured AI feedback.

Core features:

  • Video interview simulations
  • Role-specific interview questions
  • Custom interview creation
  • AI evaluation on clarity, structure, confidence, and relevance
  • Actionable feedback, not generic tips

No community.
No job board.
No fluff.

Everything else was noise.


Tech Stack (Solo-Friendly by Design)

I optimized for speed, reliability, and low maintenance.

  • Frontend: React + Tailwind

  • Backend: Supabase (auth, DB, storage)

  • AI stack:

    • Whisper for transcription
    • Gemini + GPT-4o for evaluation and scoring
    • Structured JSON outputs (non-negotiable)
  • Payments: Stripe

  • Hosting: Vercel

  • Automation: Make.com for glue logic

Key principle:

If something required babysitting, I didn’t ship it.


The Hard Part Nobody Talks About: Evaluation Logic

The hardest problem wasn’t AI generation.

It was scoring.

Users don’t trust vague feedback like:

  • “Good answer”
  • “Needs improvement”

So I had to design:

  • Clear evaluation dimensions
  • Consistent scoring logic
  • Deterministic output formats
  • Feedback users could act on

This meant:

  • Iterating prompts dozens of times
  • Forcing models into strict schemas
  • Rejecting “creative” answers from AI
  • Treating feedback like a product, not a feature

This took more time than the entire frontend.


Pricing Strategy (No Overthinking)

I didn’t A/B test endlessly.

I went with:

  • Free: limited interviews
  • Plus: serious candidates
  • Premium: heavy users and job switchers

Why subscriptions?

  • Predictable revenue
  • Clear value progression
  • Easier to optimize LTV over time

Most importantly:

Pricing filtered users who actually cared.

Support tickets go down.
Feedback quality goes up.


Launch Strategy: No Big Bang

I didn’t do a flashy launch.

I:

  • Soft-launched to existing users from my other product
  • Posted in relevant communities
  • Wrote SEO-driven content early
  • Iterated based on real usage, not opinions

Indie lesson:

Distribution beats polish every time.


What Worked

  • Building for a pain I personally had
  • Shipping fast and improving weekly
  • Saying no to “nice-to-have” features
  • Treating AI as infrastructure, not magic
  • Keeping costs extremely low

Advice If You’re Building Solo

  • Start from a problem you deeply understand
  • Keep scope painfully small
  • Ship something useful, not impressive
  • Treat AI like a system, not a shortcut
  • Charge earlier than you’re comfortable with

Solo building isn’t easy—but it’s incredibly clarifying.

on January 15, 2026