- Freshly generated questions, every session
- A real code editor — Monaco, syntax-highlighted
Practice the interview for the job you actually want.
Paste a job posting. Get original coding questions built for that exact role, then run through a real interview — approach, code, follow-ups — with an AI interviewer that scores how you think, not just whether the code runs.
- Guiding questions before you write code
- A report that finds patterns, not just scores
Live preview
interviewer> Given a stream of incoming API request timestamps, write a function that returns how many happened in the last 60 seconds.
function requestsInLastMinute(ts, now) {
return ts.filter(t => now - t <= 60000).length;
}follow-up> What happens to this if it's called on every new request as traffic scales up?
One real example, before you start
"...Looking for a Backend Engineer with 3+ years in Node.js and distributed systems. Comfortable with concurrent request handling and API design..."
▊
function requestsInLastMinute(timestamps, now) {
return timestamps.filter(t => now - t <= 60000).length;
}5-8
Original questions generated per session
3
Independent scores per question — correctness, communication, reasoning
3
Languages supported — JavaScript, Python, Java
See it in action
Every screenshot below is the real product.
Approach first
It catches the gap before you've written a line of code.
State your plan out loud or in writing. A real flaw gets a guiding question back — never the answer handed to you.
interviewer> ▊
Code + follow-ups
Follow-ups about the code you actually wrote.
Not a generic checklist. The question references your specific function, your specific choices.
function firstDuplicate(nums) {
const seen = new Set();
for (const n of nums) {
if (seen.has(n)) return n;
seen.add(n);
}
return null;
}
interviewer> ▊
Session report
Scored on how you think, not just pass/fail.
Three independent scores per question, plus patterns that only show up across the whole session.
Pattern across this session
You consistently skip stating time complexity until asked directly — worth leading with it next time.
How it works
Paste the posting
A URL or the raw text. We extract the tech stack, seniority, and focus areas.
Get your questions
5-8 original questions, a genuine mix of difficulty, built specifically for that role.
Run the interview
Approach, code, follow-ups, scored — question after question, like the real thing.
Not another question bank.
Every part of the flow is built to feel like an actual interview, not a coding quiz.
Freshly generated, every time
No question banks, no LeetCode reskins. The model reads your posting and invents 5-8 original questions tailored to the exact role — every session is a different set.
Approach before code
A real interviewer doesn't hand you a blank editor. You explain your plan first — flawed reasoning gets a guiding question, not a lecture.
A real code editor
Monaco, the engine behind VS Code. JavaScript, Python, and Java, syntax-highlighted.
Follow-ups that read your code
Complexity, scaling, edge cases — asked about the specific function you just wrote, not a generic checklist.
Speak it, don't type it
Talk through your approach out loud and hear the interviewer respond, with a genuinely human-sounding voice.
A report that finds patterns
Not just per-question scores — it flags habits repeated across the session, like skipping complexity until asked.
Pricing
FREE first session
Full flow, every question, the whole report. No card, no signup.
$12 per session after
One new job posting, one new set of questions.
FAQ
You can — but you'd have to rebuild the interview discipline yourself every time: remembering to ask for the approach first, not caving and giving away the answer when someone's stuck, tracking follow-ups tied to the actual code, and scoring consistently. Codrill enforces all of that automatically, the same way every session, so the practice is actually comparable across attempts.