How Myskillgrid decides what to ask, and how it scores you

1 Skill maps: what your job needs

Every one of the 900+ job titles has its own skill map: the 4 to 10 skills that interviews for that job really test, picked from a pool of 700+ skills across software, banking, nursing, teaching, sales, trades, government and more. A Staff Nurse is tested on patient care and medication, an ICU Nurse on emergency scenarios and vitals, an Accountant on journal entries and GST.

Up to 4 of a job's skills are core: the ones any interview for that job is expected to cover. They are asked first, and readiness can't say Interview-ready until each one is shown (see Readiness). The rest are supporting skills. Every job also has the same behavioral skills (ownership and impact, teamwork and conflict, learning and growth) and communication skills (clear explanation, structured thinking, trade-off analysis).

Scores are kept for each skill, not just for the interview as a whole. That is what makes it a grid: you can see that your OOP is strong while unit testing is still untested. A skill is the same skill in every job that has it, so if you switch from Java Developer to Java Full Stack Developer, your scores on the skills both jobs share come with you. Your exact job title also shapes the wording of each question. Find your job and its skills on the jobs page.

2 Planning: which skill each question tests

Before the interview starts, the app (not the AI) plans every question: which skill it tests and how hard it is. Each skill gets a priority from your profile:

• Not tested yet. Skills with no evidence come early, and skills your resume claims come before other untested ones, so claims are checked first.
• Core to your job. Your job's core skills get a boost, so your first interviews cover them.
• Weak or barely measured. The lower a skill's score, the higher its priority, with a boost while it has fewer than two answers.
• Slipping. A skill that dropped in your last session, or was weak in two or more sessions, moves up.
• Stressed by the job. Skills your target job description emphasises get a boost.
• No repeats. A skill already asked in this interview is pushed back, so one interview covers several skills.

The interview type limits the choice: a technical interview asks only role skills, a case or system design interview only scenario skills, a behavioral interview only behavioral ones, and a mixed interview all of them. Only the types your job has skills for are offered.

Difficulty follows your score on that skill

During the interview the plan can move: after a clearly strong answer the interviewer can make the next question one level harder, and after an answer that misses the basics, one level easier. A question you have already seen never changes.

Candidates with 3–6 years of experience never get level 1 questions. Questions are written in English that matches your experience: short sentences and common words for freshers. In untimed interviews and practice drills you can also see each question in one of 12 Indian languages, including Hindi, Telugu, Tamil, Kannada.

3 Your resume: questions about what you did

A real interviewer asks about your own projects, and so does Myskillgrid. If you add your resume, the app picks up to 5 concrete things it says: a project, a tool you used on the job, a result with a number. Each is linked to the skill it shows, when your job has that skill.

A full interview asks about two of them, as questions 2 and 4: what you personally did, why, and what happened. When the item is linked to a skill, your answer is scored on that skill, so the claim is checked, not just taken at its word. An item with no matching skill becomes an experience question in mixed and behavioral interviews. The free interview and practice drills don't ask about your resume.

Your resume and a job description are also hints for planning: claimed skills are tested early, and skills the job description stresses get a boost. They are never counted as evidence on their own; only your answers are.

4 Follow-ups: probing a weak answer

After each answer the interviewer decides, in a second or two, whether to follow up. A shallow, vague or partly wrong answer gets a follow-up straight away that quotes what you said and asks for the missing part, like a real panel would. That measures reasoning, not memorized definitions.

A full interview has 10 main questions and up to 2 follow-ups; the free interview has 3 questions and at most 1 follow-up. Follow-ups are scored like any other answer.

5 Scoring: six criteria, fixed weights

Every answer is scored from 0 to 10 on the same six criteria, for every candidate and every job. Empty, off-topic or "I don't know" answers score 0–2 on every criterion. The criteria are combined with fixed weights into an answer score out of 100:

A worked example

A fresher is asked: "Your Spring Boot API returns 200 OK even when the order doesn't exist. How would you fix it?" They answer that they would throw an exception and handle it in a @ControllerAdvice so the API returns 404. Correct and practical, but it doesn't weigh alternatives.

The lowest criterion here is trade-offs, so a natural follow-up is: "Would you return 404 or an empty list when a search finds nothing? Why?" With every score you also get what went well, what to improve, why the skill matters in this job, one thing to practice, and a stronger model answer. You can retry any question from the report.

6 From answers to skill scores

One answer often shows more than one skill. The evaluator lists the 1–3 skills of your job that the answer gave clear evidence about, each with a score out of 100 and a one-line note quoting or paraphrasing what you said. Each piece of evidence gets a weight:

Recent answers count more. A skill's score is a weighted average of its evidence, and every newer answer on that skill shrinks the weight of older ones by × 0.75. If you scored 40, then 60, then 75 on a skill, your score is 62, not the plain average of 58, because you have been improving.

Only fair evidence counts. The evaluator also checks that the question really tested the skill it was meant to. If it didn't, your answer counts for the skills it did show, not for that one. A spoken answer we couldn't hear is left unscored, never given a low score.

Enough evidence. A skill counts towards readiness once its evidence adds up to 0.8: one direct answer is enough, a single passing mention is not. Confidence is high from 2.0. Every score links to the answers behind it, so you can see why it is what it is.

7 Skill status: strong, needs attention, weak

Strong has to be earned on the skill itself. A skill is called strong only after a question on that skill at level 2 (applied) or harder. Until then a high score reads Strong on basics (only easy questions so far) or Strong so far (only mentioned in answers about other skills), until an applied or harder question on that skill confirms it. A score resting on too little evidence also reads Strong so far: communication skills, measured from every answer, need a few answers before they can be strong.

Resume claims are checked. A skill your resume claims shows as unverified until an interview has enough evidence, then as verified (score 55+) or not confirmed.

8 Readiness: one number for the whole job

Interview readiness is the average score of the skills with enough evidence. Skills your target job description stresses count 2 times, because they matter more in that interview. Readiness comes with coverage: how many of the job's skills are assessed so far. The job's key skills (its core skills, and any your job description stresses) must each be shown on their own before readiness can say Interview-ready: an average can hide a key skill nobody has asked about yet. Communication is not part of readiness: it already counts in every answer's score, so adding it again would count it twice. It is shown on its own instead.

Your dashboard shows readiness after every session as a line chart, and your skill grid with every skill colored by status, so you can see whether practice is working. You can also share a snapshot of your readiness and top 3 skills as a skill card link.

9 Practice plan and drills

The practice plan lists up to three skills to work on next, weakest first, each with a reason you can check: "Weak in 2 sessions", "Dropped 8 points last session", "On your resume, not yet verified" or "Important in your target job".

A practice drill is 3 questions on one skill. It starts at the level your score calls for and climbs a level with each question. After each answer the interviewer can also make the next question a level harder if the answer was clearly strong, or a level easier if the basics were missing, never more than one level from the plan. Weak skills stay near the top of the plan for your next full interview, so it tests them again, and each skill's history shows what changed and why.

10 Where AI is used, and where it isn't

Your answers are treated only as answers: anything in them that looks like an instruction to the AI is ignored.

11 Limits

• An AI evaluator can misjudge an answer. Fixed criteria and weights keep scores consistent, and every score shows the answer behind it, so you can see when it got something wrong.
• Readiness measures practice answers on your job's skills. It does not predict hiring or the result of a real interview, which depend on many things no tool can see.
• Spoken answers are transcribed first; a transcription slip such as "hash map" for HashMap is not counted against you.
• Your answers and resume are visible only to you. If you join an institute's batch, its staff see your readiness and scores, never your answers or resume. See the privacy policy.

Questions