An AI interviewer that speaks out loud
BladeSkill runs a voice mock interview, gives feedback on every answer, finds the gaps in your knowledge and shows how well prepared you are for your next interview.
Free plan
No card required
Three languages
AI interviewer
React Frontend Developer · Middle
Interview in progress
TOPIC 3 OF 6
What happens if you update state inside useEffect without a dependency array?
Interviewer speaking
Your answer — by voice or text
8.6
/ 10
ANSWER SCORE
Core idea is right. Missing the render queue — add that and the topic is closed.
READINESS
72%
4 500+
questions in the bank
1 600+
coding tasks
20
career tracks
3
content languages
These numbers are computed from the database at build time, not typed in by hand.
Not "grind some questions and good luck" — a measurable picture of your own preparation.
Pick your track
20 specializations, from React and Python to DevOps and QA. From there the platform picks what you should work on and hides the rest.
Train in four modes
Voice interviews, free-text answers scored by AI, quizzes and live code in the browser. One shared progress, wherever you answer.
Track readiness, not a counter
The skill radar finds your weak topics, and the AI report turns them into a list of what to do next.
Each mode trains a different skill: explaining out loud, writing it in your own words, recognising it among options, and writing the code.
The real differentiator
The interviewer asks out loud, listens to your answer, asks follow-ups and works through anchor topics from the bank. It ends with a summary of every topic you covered.
Depth check
Answer in your own words, the way you would in a real interview. The AI scores the technical depth and tells you exactly what was missing.
Fast check
The AI generates options for each specific question, and the correct one never reaches the browser until the quiz is finished — you cannot read it off the page.
Hands on
JavaScript and TypeScript tasks. Code runs right in the browser, tests check the result, and the AI reviews the solution itself — not just pass or fail.
One progress across every mode
Every answer — in an interview, a quiz, the trainer or a coding task — lands in the same history. Your skill radar, review schedule, level and readiness are all computed from it.
CHAPTER 2
React: state and effects
2 / 4
State hooks
Done · 18 of 18 questions
useState tasks
Done · 6 of 6 tasks
Boss: effects and subscriptions
Current · 7-question quiz
Memoisation and wasted renders
Unlocks after the boss
Chapters by difficulty, each holding topic nodes. A node closes only when the theory is done and the boss quiz is passed; a chapter closes when every node is.
Instead of choosing between five sections, the platform decides what helps more right now — revisiting something fading, or moving on.
You cannot close a topic by scrolling past the theory: it takes a quiz with a passing threshold.
Everything sits together: read it, apply it, check it. Not three separate sections you have to bridge yourself.
Every answer sharpens the map of your knowledge: what stuck, what you only recognise among options, and what is due for review.
A breakdown by specific topics rather than one overall percentage. You can see "closures — 2 out of 10".
An SM-2 schedule brings back what you are forgetting exactly when it starts to fade.
One number you track for yourself, with the reasoning behind it: why that figure, and what would raise it fastest.
It sees your screen and your code, explains, nudges, and keeps you from getting stuck on a task.
SKILL RADAR
React Frontend Developer
72%
readiness
Hooks
86%
Async
64%
Performance
41%
Closures
22%
An example breakdown. Yours is built from your own answers, starting with the very first question.
Preparation fails not because it is hard, but because people drop it on day three.
Duels
Challenge a friend to a question battle
One question set for two players, scored by correct answers and then by time. Answers still count toward your progress — a duel is not a side game, it is the same training with an opponent.
Ghost
Asynchronous, over a link: your opponent plays whenever they like and you watch their pace round by round.
Live
A lobby, a synchronised start, and both of you watching the other's progress as answers land.
Every answer earns XP, and each level raises your limits and hands you a wheel spin.
Counted from real activity. A missed day can be covered with a freeze won on the wheel.
Three tasks a day, deterministic — two browser tabs will show you exactly the same ones.
For volume, for individual modes and for duel wins. Awarded automatically.
One global XP standing for the week, a live leaderboard and rewards for the top three on Monday.
A free spin every day, plus spins for levels and goals. The prizes are real: interviews, streak freezes, premium.
Invite a friend
Your invite link lives in your profile. Once your friend finishes their first steps, you both get spins.
20 specializations with the real number of questions behind each. Pick yours and the platform filters out the rest.
React Frontend Developer
2489 questions
Python Backend Developer
702 questions
Node.js Backend Developer
657 questions
Java Backend Developer
602 questions
Golang Backend Developer
584 questions
C# Backend Developer
579 questions
C/C++ Backend Developer
498 questions
DevOps Engineer
488 questions
PHP Backend Developer
476 questions
Ruby Backend Developer
455 questions
Rust Backend Developer
454 questions
Machine Learning Engineer
439 questions
The short answers to what people ask before signing up.
Yes. The question bank, quizzes, the trainer, coding tasks and the voice interview all work on the free plan — only daily limits apply. No card is needed to sign up.
Yes. The interviewer speaks the questions out loud and you reply by voice or text. It asks follow-ups based on your answer and finishes with a summary of every topic you covered.
English, Ukrainian and Russian — the interface, the questions and the interview voice. You can switch language in the site header or in the app settings.
React and frontend, Node.js, Python, Java, Go, C#, C/C++, PHP, Ruby, Rust, DevOps, Machine Learning, Data Science, Android, iOS, QA, embedded systems, game development and more. Frontend has the deepest coverage.
Higher daily limits on AI operations and interviews, plus a higher-quality interviewer voice. None of the core modes are locked behind the paywall.
They build your personal history, which is what your progress, skill radar and review schedule are computed from. The text of your answers is never published or shown to other users. Your results stay yours: BladeSkill is a self-study tool — we do not screen candidates for employers, never share your results with recruiters, and the platform makes no decision about you.