INTERVIEW PRACTICE FOR THE AI-NATIVE ERA

Practice AI-native
coding interviews

Code with an AI assistant, explain your decisions to a live interviewer, and get a timestamped report on how you actually performed.

Start free interview See how it works

Backend Interview / Build a URL Shortener API

27:14 remaining5/7 passingAI patch proposed

shortener.py

1 from fastapi import HTTPException
2 async def create_short_url(payload):
3     if not payload.url.startswith(("http://", "https://")):
4         raise HTTPException(422, "Invalid URL")
5     key = await generate_key()
6     async with repository.lock(key):
7         await repository.save(key, payload.url)
8     return {"short_url": f"/{key}"}

10 # Candidate owns final review

VISIBLE TESTS✓ creates a short URL✓ rejects malformed URL✕ prevents duplicate key writesExpected 409, received 500 · race under concurrent requests

Review AI patchAI suggests a change. Review the diff before it lands.

01 / CHOOSE

Choose realistic backend tasks

Practice on small repo-style backend tasks, not toy whiteboard problems.

FastAPI + pytestURL shortener API30 min

Fastify + VitestInventory service30 min

Django RestTeam invitation flow35 min

Express implementationRate-limit middleware30 min

Auth defectsSession refresh route30 min

Edge casesOrder invariant suite35 min

02 / COLLABORATE

Code with AI, but stay accountable

Use AI like you would in a real AI-forward interview. The platform tracks whether you reviewed, tested, and understood the changes.

AI coding assistant · Candidate-controlled

I found a validation race that can create duplicate keys under concurrency.

PROPOSED PATCH- await repository.save(key, url)+ async with repository.lock(key):+ await repository.save(key, url)

03 / DEFEND

Defend your decisions out loud

A live voice interviewer follows the work, applies time pressure, and asks for reasoning—not narration.

Voice interviewer● LIVE

“Walk me through why this patch is safer under concurrency.”

YOUR ANSWER

The lock makes key generation and persistence one atomic decision...

04 / IMPROVE

Get an evidence-based report

Every score is tied to hidden tests, patch decisions, debugging behavior, and timestamped voice evidence.

Performance reportSession · URL Shortener API

PERFORMANCE SCORE82Strong ownership

You reviewed AI output, caught a contract regression, and explained the final tradeoff.

Correctness

AI collaboration

Prompting quality

Code review judgment

Debugging

Communication

Time management

Overreliance risk

Timeline evidence