Written up by candidates straight out of the interview, then filed by company and round, and dated wherever the report gave a month.
Free to start · no card needed
Updated as candidates report new questions.
Questions from
Every question here was asked in a real interview.
Written up right after the loop, then checked against other reports of the same round.
A question carries the month it was asked wherever the report gave one, and the list leads with the recent ones and the ones the most write-ups cite.
Coding, SQL, system design, object design and AI pairing, each in its own workspace.
Candidates post how their rounds went, in order. Read one before you take our word for any of this.
Design a rate limiter for a public API. Start on one node, then make it hold across a fleet.
Every question reported for that loop, in one list.
Interview reports and the questions they came from, in one dataset — every round of a big-tech loop, checked by hand by engineers from those companies. The first few at any company are free.
| Feature | TrueInterviewEnd to end | Leet(name partly hidden)Question bank | 1poi(name partly hidden)Interview forum | InterDB(name partly hidden)Reports site |
|---|---|---|---|---|
| Built for big-tech loops | Yes | No | Yes | No |
| Reports open the question | Yes | No | No | No |
| Verified by big-tech engineers | Yes | No | No | No |
| Every round, not just coding | Yes | No | No | No |
| Judged on hidden tests | Yes | Yes | No | No |
Practising is the half you do. The account does the paperwork around it, right through to the submit button.
Messages people sent us, and the write-ups they post on the board.
The question bank has enough in it now that I can actually prep off it. I'm using it while I interview at other places.
I'm interning right now and want to go quant next. That's what I'm using it to prep for.
The content holds up. It's accurate and the quality is good.
I spent two years job hunting at big tech companies. I interviewed at Databricks, Stripe, Optiver and Uber, and took engineering offers from Amazon and Philips.
What I worked out along the way is what this site does, from tailoring a CV for one posting to negotiating the offer at the end. The rubric that scores your practice answers is the one I scored my own against.
Preparing for Stripe I kept meeting the same questions, at company after company. So I built the crawler behind the bank: it pulls interview reports from several sources, merges them, and ranks a question by how often candidates report it.
I still practise on the other platforms and talk to engineers who have just changed jobs. That is what decides which part I build next.
OptiverUber
PhilipsCandidates who just sat the interview. Some post on our board, which anyone can read; the rest come from public reports we verify and file by company. A question shows the month it was asked wherever the report gave one, so an old one says so instead of being presented as current. The full sourcing and verification process is at /methodology.
No. System design, object and low-level design, SQL and the AI-pairing rounds are all here alongside the algorithm rounds, each in the workspace that round calls for: the whiteboard for system design, a query workspace that runs against real Postgres for SQL, a multi-file project for AI coding.
The default view at /problems is a directory of companies rather than a list of questions. One row each, showing how many questions we hold for that company and how many formats they cover, with a search box for typing a name. A company appears once candidates have reported questions from it, and two spellings of one employer collapse into a single row. A New marker means every dated report we hold from that company is inside the last six months.
LeetCode is a puzzle library. This is a record of what specific companies asked, across coding, SQL, system design, object design and the AI-pairing rounds.
Python 3, JavaScript, Java, Go, C++ and C, each with a starter and a real formatter. SQL questions run PostgreSQL rather than a cut-down engine, so window functions and CTEs behave the way they will in the interview. Coding and SQL are judged on our servers, never in your browser.
A reference solution comes with the question, and coding and SQL are judged server-side on hidden tests. When you fail a case you get to read that case, rather than a red cross with nothing behind it. On a coding question that means every case you failed, with its input, the expected output and what your program printed. SQL opens the first dataset your query got wrong, with its rows and the table you were meant to return. A hidden case you passed shows a status and nothing else, so a working solution never enumerates the suite.
It finds roles that fit you, writes a CV and cover letter for the ones worth applying to, fills in the application and sends it once you have read it and said yes. The CV is rewritten against the posting itself rather than the job title, and it invents nothing you never listed. Every application lands on your tracker and stays current to the offer.
No account to start: open a free question and run the tests, in the same workspace and against the same judging a paid one gets. An account saves your progress across devices. Pro opens the rest of the catalog and Insider adds every company's bank. What you pay for is coverage and the AI review, not access to a working product.
Two products, and Insider contains Pro. Pro opens the TrueInterview 75 and 150 question sets, the Career Co-pilot, per-role CV tailoring and the AI review of a finished solution. Insider adds every company question bank, reported by candidates who sat those rounds. Either one lifts the daily read limit on the interview write-up board. The TrueInterview 20 and the first few questions at every company stay free on both. Buy Insider if you already know where you are interviewing. Both cards are at /premium.
In US dollars. The plan registry prices everything in USD, and Stripe holds one USD price per product per cadence. The cadences are lifetime, yearly, quarterly and monthly. Quarterly and yearly take a single payment up front for the whole period, so Insider yearly is one charge of $159 covering 12 months, and again a year later. Lifetime is one charge of $359 that never renews. The cards show the effective monthly rate first and print the real charge beside it. That charge is the number your card sees. Prices are at /premium.
No. Your work is sent to the model providers named in the Privacy Policy — OpenRouter, which routes to DeepSeek models, and Anthropic — to do the job you asked for, and under the API terms we use your inputs are not used to train their models. Non-essential analytics and advertising cookies also stay off until you accept them on your first visit. Export or deletion requests go to kevin@trueinterview.io and are handled under the Privacy Policy, and your CVs live in a document library you control, each version kept.
Every round they will run, from the people who sat them, judged against the real tests.