Walk into the interview having done it already.

Alex reads your resume and the job, asks what that role would ask, pushes back, then shows you where you lost the room.

Start a free mock interview Read a sample report

Free · In your browser · Nothing recorded unless you choose

Report, behavioral round

HireAt the Regular bar

Skills, 1 to 5

Specificity and ownership3 of 5
Structure4 of 5
Impact and results4 of 5
Self-awareness and learning4 of 5
Collaboration and communication3 of 5

I think he said the catalog page was going to be slow because it, uh, loads all the sections.Your words, 20:05

Say this instead

Marco’s position was that the catalog page loads all the sections, so it would be slow for users and caching would fix it. I disagreed, but I should be able to state Marco’s case in one sentence.

Moment, 8:30: You hesitated: 5.6 s to start, your usual 3.0 s
Voice signal: Pitch at phrase ends: went up at 18 of 319 measured phrase endings

Sample data, fictional student

Read the full sample report

Report, behavioral round

HireAt the Regular bar

Skills, 1 to 5

Specificity and ownership3 of 5
Structure4 of 5
Impact and results4 of 5
Self-awareness and learning4 of 5
Collaboration and communication3 of 5

I think he said the catalog page was going to be slow because it, uh, loads all the sections.Your words, 20:05

Say this instead

Marco’s position was that the catalog page loads all the sections, so it would be slow for users and caching would fix it. I disagreed, but I should be able to state Marco’s case in one sentence.

Moment, 8:30: You hesitated: 5.6 s to start, your usual 3.0 s
Voice signal: Pitch at phrase ends: went up at 18 of 319 measured phrase endings

Sample data, fictional student

Read the full sample report

Built for students at universities everywhere

Directions

Software Engineering
Coding, language depth, system design
Data Science
SQL and pandas, statistics and experiments, metrics design
Machine Learning
ML coding, ML theory, ML system design
Quant Developer
Low-latency C++, probability and mental math, trading systems

Languages

Python
Java
Go
C++
TypeScript

Software engineering in any of the five. Quant in C++. Data science codes in SQL and pandas.

Pick your direction

Your round. Not a generic one.

The round where you code out loud.

Write it, run it, defend it. Alex asks about the case that failed.

merge.pyCoding, Python

Given a list of intervals with start and end times, merge all overlapping intervals and return the result.

def merge(intervals):

intervals.sort(key=lambda iv: iv[0])

merged = []

for start, end in intervals:

if merged and start <= merged[-1][1]:

merged[-1][1] = end merged[-1][1] = max(merged[-1][1], end)

else:

merged.append([start, end])

return merged

Example 1
Edge: touching intervals
Edge: unsorted, one contains the others
Edge: a single point interval

Run examples

3 of 4 examples pass

Edge: unsorted, one contains the others: expected 1 10, got 1 3, 5 6

4 of 4 examples pass

Including the case that failed

Alex

Before you change anything: what’s happening on that input? Walk me through it.
The four examples are the ones I gave you. What else would you test?

Your report would say

“Your merge overwrote the end instead of taking the max, so a containing interval lost its end and the third example failed.”

Next: Trace edge cases before you run20 min

Write merge intervals from scratch, hand-trace the contained, touching and identical cases on paper first, then run and compare against your traces.

See the full sample

The round where the join goes wrong.

Real SQL on real tables. Alex asks what every row means.

retention.sqlSQL on DuckDB

Calculate day-1 retention: the share of users who logged in on the day after their first login, for each signup date.

WITH first_login AS (

SELECT user_id, MIN(login_date) AS cohort_date

FROM logins GROUP BY user_id

)

SELECT f.cohort_date,

COUNT(DISTINCT f.user_id) AS cohort_size,

COUNT(DISTINCT l.user_id) * 1.0

/ COUNT(DISTINCT f.user_id) AS day1_retention

FROM first_login f

JOIN logins lLEFT JOIN logins l

ON l.user_id = f.user_id

AND l.login_date = f.cohort_date + 1

GROUP BY f.cohort_date ORDER BY f.cohort_date;

cohort_date	cohort_size	day1_retention
2024-01-01	1	1.0
2024-01-02	1	1.0

cohort_date	cohort_size	day1_retention
2024-01-01	2	0.5
2024-01-02	2	0.5
2024-01-03	1	0.0

Run examples

0 of 1 examples pass

Example 1: 3 rows expected, 2 returned

3 of 3 examples pass

Alex

Walk me through what the left join produces for user 4, the one who never came back.
Why COUNT DISTINCT on the logins side?

Your report would say

“You hand-computed expected results before writing code and used failing comparisons to isolate bugs yourself in all three questions.”

Next: Commit to a default, then caveat20 min

Answer three open-ended prompts out loud in 60 seconds each ... always starting with ‘My default would be...’ before any caveat.

See the full sample

The round where accuracy is a trap.

Theory with real numbers. Alex pushes until the metric fits the problem.

Fraud detector, 0.1% fraudML theory, no code

You are building a fraud detector where 0.1% of transactions are fraud. How do you evaluate it?

The metric you pick

ROC AUCPR AUC

“…a ranking metric that doesn’t depend on the threshold, like ROC AUC.”

“Precision and recall are better here: PR AUC, and precision at the recall the business needs.”

One day of alerts

Flagged, fraud

950

Flagged, not fraud

Missed fraud

the rest

Not flagged, not fraud

Precision: 50 / 1,0005%
Recall: 50 / 8062.5%

Alex

Is ROC AUC a good fit at 0.1%?
Your model flags a thousand transactions a day. Fifty of them are fraud, and there are eighty frauds a day in total. What are precision and recall?

Your report would say

“Evaluation reasoning is genuinely strong: you rejected accuracy, then rejected ROC AUC at 0.1% positives, and moved to PR-AUC, cost-based thresholds and calibration without being told the answer.”

Next: Lead with the mechanism, not the history20 min

Take 5 ML questions you know well and answer each in 60 seconds, mechanism first, history only if asked.

See the full sample

The round where the clock is running.

Mental math out loud, on the clock. Alex changes the rules until a shortcut breaks.

Problem 2 of 5Probability, no paper

You roll a fair die and are paid the number shown in dollars. You may roll once more and take the second result instead. What is the game worth?

Round clock

22:24

of 30:00

Mental math, said out loud

a fresh roll, worth 3.5 on average
keep 4, 5 or 6, and reroll 1, 2 or 3
2.5 plus 1.75, 4.25 dollars
2.5 plus half of 4.25, so 4.625
I only keep a 5 or a 6
28 sixths, about 4.67

Alex

What if you may reroll up to two times?
Is a 4 worth keeping when rolling on is worth 4.25?

Your report would say

“Two answers started with a quick wrong answer the interviewer had to nudge you past: the 1/4 = 4 flips guess on the coin problem and reusing the 3.5 threshold on the two-reroll game.”

Next: Set the threshold to the value of continuing20 min

Run the die game with three rerolls and with a one-dollar cost per reroll. Each time, state the value of continuing first, then derive the keep threshold and the game value.

See the full sample

The product

From your resume to your next round.

01Prepare
Your resume becomes your plan.
Upload your resume and the job posting. Check what Alex understood, then get questions for every round. Behavioral ones name the project or job line they came from.
Sample preparation: a confirmed resume with projects and skills, all seven preparation steps done, where you match the role and the topics to expect.
Sample preparation: a confirmed resume with projects and skills, all seven preparation steps done, where you match the role and the topics to expect.
02Interview
Talk it through, out loud.
Alex asks, you answer. The question stays on screen, a timer keeps the pace, and your code or notes sit right next to it.
Sample live interview: Alex asks short questions, the candidate answers, a timer and question progress sit in the top bar, notes in the workspace.
Sample live interview: Alex asks short questions, the candidate answers, a timer and question progress sit in the top bar, notes in the workspace.
03Report
Know exactly what to fix.
A score, a short read on how it went, and the one change that moves it most. Every point quotes what you said, plus how you delivered it: pace, pauses, fillers and hedging.
Sample report: score 76, hire, at the bar, three points with the student’s own words, and one next task: a reflection drill.
Sample report: score 76, hire, at the bar, three points with the student’s own words, and one next task: a reflection drill.
04Progress
See the curve bend.
Scores over time, your streak, and how ready you are for each round, so you know where the next hour goes. Close to the real thing? Run the whole loop for one role, round by round, with a debrief at the end.
Sample progress: a four-day streak, a score line rising from 52 to 76, and readiness bars by round with system design weakest.
Sample progress: a four-day streak, a score line rising from 52 to 76, and readiness bars by round with system design weakest.

Illustrative screens with demo data.

How you are scored

Scores with receipts. And how you sounded.

What a behavioral round is scored on, 1 to 5 each, against the bar for your level:

Specificity and ownership
Structure
Impact and results
Self-awareness and learning
Collaboration and communication

Quotes with the time: Every point links to what you said and when, so you can find it in the transcript.
One thing to do next: The report ends with a single task and a drill, not twenty tips.
Signals, not verdicts: Pace, pauses, fillers and hedging are shown against a comfortable range, never as a judgment of you.

Read a full sample report

How you sounded

Pace146 wpm
Steady across most answers.
Before you start3.2 s
About three seconds to think on average; two questions took 6 to 7 s.
Filler words2.1 / min
Mostly steady, with one burst of 19 a minute on a single answer.
Hedging35 phrases
“I guess”, “kind of”, “I think”, spread across the interview.
Endings0 of 49 faded
Your volume held at the end of every measured answer.

Where it changed: on the conflict story you sped up to 184 words per minute, with 19 fillers a minute (your usual 2).

Your interviewer, Alex

A chatbot says you nailed it. Alex asks why.

Real interviewers do not cheer you on. They push, wait and write everything down. So does Alex.

No cheerleading.: Alex stays neutral: no praise, and no answer or hint you did not ask for.
A real clock.: Timed rounds with a warm-up, core questions and a wrap-up, so you learn to pace yourself.
Your answers, quoted.: The report quotes what you said and ends with the one thing to fix.

Sample coding round, top k frequent numbers

Youm can be as big as n, so O(n log n) worst case. Space O(m) for the counts.
A chatbotSpot on, great analysis! Want to move on to the next question?
Alex
Think about what you're sorting. You only need the top k, not a full ordering. What structure exploits that?
YouA heap. A min-heap of size k, keyed on frequency.
Alex
When does the difference between log m and log k actually matter?
YouWhen k is much smaller than m.
Alex
Walk me through the heap on the first example before you write it.

Read this round in full

System design

Draw it. Alex is watching.

You draw on a shared whiteboard. Alex watches every block you add and asks about it, the way a real interviewer would. Pick your direction and drag through a real session.

See a system design report

mockich / system designShared whiteboard

Design a distributed rate limiting service used by hundreds of internal APIs.

Alex’s board notesRules DB looks like a single instance on the main path. Worth asking about.

Alexsees your board as you draw

Start

Coverage

Every round they throw at you.

Drill one round, or run the whole loop for a role. Later rounds remember what you said in the earlier ones.

Questions built from your resume and the job description, drawn from a curated bank of 1,000+ questions for every track, level and round.

Recruiter screen
Your story, why this role, the logistics. Fifteen minutes that decide if you get the rest.Full loop
12 to 15 min
Coding
Real problems in a real editor. Your code runs, and Alex asks why. See a sample coding report
45 min
Fundamentals
The “do you actually know this” round: language internals, CS basics, stats or ML theory. See a sample fundamentals report
35 min
Probability
Mental math and probability puzzles, against the clock.Quant See a sample probability report
30 min
System design
Draw the system, then defend it when Alex asks what breaks at ten times the load. See a sample system design report
45 to 50 min
Behavioral
Conflict, ownership, failure. The questions come from your own resume. See a sample behavioral report
30 to 35 min
Hiring manager
One project, taken apart. Then: why this role?Full loop
25 to 30 min

Two levels: regular, and lead for senior roles.

Private by design.

No recordings by default
Your voice runs the interview and is not kept unless you switch that on. Saved recordings go after 90 days.
Never sold
Not sold, not used for ads, never shared with employers.
Kept for a year
Transcripts and reports are deleted after 365 days unless you keep them longer.
Yours to take or delete
Download everything as one file, or delete it all, from Settings.

Before you start.

How is this different from a chatbot’s voice mode?

A chat assistant is built to be agreeable: it cheers you on and moves along. Alex is built to interview. Alex stays neutral, with no praise and no hint you did not ask for, runs a real clock for each round, and follows up on what you actually said. Then you get a report that quotes you.

Who interviews me?

Alex, an AI interviewer. Not a person, and not pretending to be. Alex reads your resume and the job posting, asks what that role would ask, follows up and takes the notes your report is built from. A friend who has hired before is still worth a lot; Mockich is for all the rounds before that one, whenever you want.

How long is a round, and can I do just one?

Rounds run from 12 to 50 minutes: a recruiter screen is 12 to 15, behavioral 30 to 35, coding 45, system design 45 to 50. Do a single round, or the whole loop for one role across several days. If you refresh the page or lose your connection mid-interview, you pick up where you were.

What do I need?

A browser and a microphone. Headphones help. Bring your resume and a job posting for questions about you, or just start a single round. Interviews are in English, for software engineering, data science, machine learning and quant roles, at regular or lead level.

What if I freeze or get nervous?

Good, that is what this is for. Take your time, say you are thinking, or type your answer. The report shows where you hesitated, so you know what to rehearse.

What do I get at the end?

A score against the bar for your level, a short read on how it went, and points that quote what you said with the time you said it. It ends with the one thing to fix next, and a drill for it. Voice rounds also show your pace, pauses, filler words and hedging, as signals against your own usual, not verdicts.

Is my voice stored?

Not by default. Your voice is streamed to our AI service providers to run the interviewer while you talk and is not kept unless you switch on “save my recordings”. If you do, recordings are deleted after 90 days, when you delete the session, or as soon as you switch it off.

Can I delete my data?

Yes. In Settings you can download everything in your account as one file, delete a single interview, or delete your account and everything in it.

Does a good score mean I will get the job?

No. It shows how this practice went against the bar for your level. Companies decide differently; the point is to walk in prepared.

Walk into the interview having done it already.

Directions

Languages

Your round. Not a generic one.

The round where you code out loud.

From your resume to your next round.

Your resume becomes your plan.

Talk it through, out loud.

Know exactly what to fix.

See the curve bend.

Scores with receipts. And how you sounded.

A chatbot says you nailed it. Alex asks why.

Draw it. Alex is watching.

Every round they throw at you.

Recruiter screen

Coding

Fundamentals

Probability

System design

Behavioral

Hiring manager

Private by design.

No recordings by default

Never sold

Kept for a year

Yours to take or delete

Before you start.

How is this different from a chatbot’s voice mode?

Who interviews me?

How long is a round, and can I do just one?

What do I need?

What if I freeze or get nervous?

What do I get at the end?

Is my voice stored?

Can I delete my data?

Does a good score mean I will get the job?