Interview rubric
How Coding Interviews Are Scored: Rubric by Level (2026)
A coding interview is scored on 5 dimensions: problem understanding (15%), approach and optimisation (25%), code correctness and quality (25%), testing and verification (15%), communication (20%). Each is rated 1 to 4 against anchors that rise with the level. At mid level, an outstanding score on approach and optimisation means: "Compares two or more approaches and chooses one with a stated reason."
The dimensions and their weights
| Dimension | Weight | Solid (3) at mid level |
|---|---|---|
| Approach and optimisation | 25% | Reaches optimal with at most one light hint and states complexity correctly. |
| Code correctness and quality | 25% | Writes clean, mostly correct code with sensible names and structure; self-corrects small slips. |
| Communication | 20% | Thinks aloud and explains the reasons behind each decision. |
| Problem understanding | 15% | Clarifies inputs, constraints and edge cases unprompted and states assumptions before coding. |
| Testing and verification | 15% | Walks through edge cases unprompted and fixes own bugs. |
What changes between levels
On approach and optimisation, the heaviest dimension, an outstanding mid-level answer "compares two or more approaches and chooses one with a stated reason", while an outstanding senior answer "reaches optimal unaided and explains when each alternative approach would be the better choice". The same answer can score 4 at mid level and 2 at senior: senior candidates are expected to raise things unprompted that mid-level candidates may be asked about.
Every score, at every level
Problem understanding (15%)
| Level | 1 · Poor | 2 · Borderline | 3 · Solid | 4 · Outstanding |
|---|---|---|---|---|
| Intern / new grad | Starts coding on a wrong reading of the problem and does not notice when it fails. | Understands the problem only after the interviewer re-explains it with an example. | Asks one or two basic questions when prompted and works through an example that matches the problem. | Asks about input size and empty input unprompted and solves the problem as stated. |
| Junior | Codes on a wrong reading of the problem and needs correcting mid-solution. | Clarifies only after the interviewer corrects a wrong assumption. | Asks basic clarifying questions when prompted and states an assumption before coding. | Clarifies inputs, constraints and edge cases unprompted before writing any code. |
| Mid-level | Starts coding on a wrong understanding and has to restart after correction. | Asks clarifying questions only when prompted and misses key constraints such as input size. | Clarifies inputs, constraints and edge cases unprompted and states assumptions before coding. | Also surfaces a hidden constraint or ambiguity and explains how it changes the approach. |
| Senior | Misreads the problem and keeps going after the interviewer signals a mismatch. | Clarifies inputs unprompted but misses the constraints that decide the approach until prompted. | Reframes the problem, states assumptions and explains how each one affects the approach. | Spots a hidden constraint the interviewer never mentioned and adjusts the plan before coding. |
| Staff | Misreads the problem and does not check understanding even after a wrong first attempt. | Clarifies the basics but needs prompting to consider scale or input guarantees. | Reframes the problem, states assumptions and ties them to scale and real-world use. | Raises constraints a real deployment would impose, such as streaming input, and adapts the plan. |
Approach and optimisation (25%)
| Level | 1 · Poor | 2 · Borderline | 3 · Solid | 4 · Outstanding |
|---|---|---|---|---|
| Intern / new grad | Cannot describe any working approach, even after several hints. | Reaches a brute-force approach only with heavy, step-by-step hints. | Reaches a working brute force, then improves it with two or three hints. | Reaches the optimal approach with one or two hints and states its time complexity. |
| Junior | Has no viable approach, even after hints. | Reaches brute force but cannot improve it despite hints. | States brute force first, then reaches optimal with up to two hints; complexity mostly right. | Reaches optimal with one light hint and states time and space complexity correctly. |
| Mid-level | Has no viable approach, or one that is wrong for the stated constraints. | Stays at brute force, or reaches optimal only with heavy hints. | Reaches optimal with at most one light hint and states complexity correctly. | Compares two or more approaches and chooses one with a stated reason. |
| Senior | Has no viable approach, or presents brute force as the final answer. | Reaches optimal only after a hint or two; the complexity statement is imprecise. | Reaches optimal unaided, compares at least two approaches and states complexity precisely. | Reaches optimal unaided and explains when each alternative approach would be the better choice. |
| Staff | Has no viable approach, or cannot explain why the chosen approach works. | Needs a hint to reach optimal, or cannot explain why it is optimal. | Reaches optimal unaided and articulates trade-offs such as memory against speed. | Reaches optimal unaided, weighs trade-offs and adds production concerns such as scale or concurrency. |
Code correctness and quality (25%)
| Level | 1 · Poor | 2 · Borderline | 3 · Solid | 4 · Outstanding |
|---|---|---|---|---|
| Intern / new grad | Code does not run or stays as pseudocode, even with help. | Code works only after the interviewer points out several bugs line by line. | Writes mostly correct code with a few bugs, fixed once the interviewer points at the area. | Writes correct, readable code with clear names and fixes own small bugs. |
| Junior | Code does not run or is badly wrong. | Code works only with major interviewer help on the logic. | Writes mostly correct code with readable names; some bugs fixed with light help. | Writes clean, correct code with sensible helper functions and few bugs. |
| Mid-level | Code does not run or is badly wrong. | Code works only after major interviewer help. | Writes clean, mostly correct code with sensible names and structure; self-corrects small slips. | Writes idiomatic, modular code with no bugs found. |
| Senior | Code does not run or is badly wrong. | Code works but is messy, with duplication, unclear names or bugs needing interviewer help. | Writes idiomatic, readable code with sensible abstractions and no bugs needing interviewer help. | Writes idiomatic, modular, bug-free code that would pass a code review with minor comments. |
| Staff | Code does not run or is badly wrong. | Code works but has structural problems or bugs the interviewer has to point out. | Writes idiomatic, well-structured code with clear abstractions and no interviewer-found bugs. | Writes production-ready code that handles invalid input and is easy to extend. |
Testing and verification (15%)
| Level | 1 · Poor | 2 · Borderline | 3 · Solid | 4 · Outstanding |
|---|---|---|---|---|
| Intern / new grad | Never tests the code, even when asked. | Traces one example only when the interviewer walks through it with them. | Traces the happy path when asked and fixes a bug the trace reveals. | Traces an example unprompted and checks one edge case such as empty input. |
| Junior | Does no testing at all, even when asked. | Tests the happy path only after being asked more than once. | Tests the happy path when asked and fixes what the test reveals. | Tests edge cases unprompted and finds at least one own bug. |
| Mid-level | Does no testing at all, even when asked. | Tests the happy path only, and only when asked. | Walks through edge cases unprompted and fixes own bugs. | Designs tests systematically, states precise complexity and catches a subtle bug. |
| Senior | Does no testing, or declares the code done without checking. | Tests edge cases only when prompted. | Runs systematic tests unprompted, finds own bugs and states precise complexity. | Designs a minimal covering set of tests and catches a subtle bug before the interviewer does. |
| Staff | Does no testing, or declares the code done without checking. | Tests some edge cases unprompted but misses whole classes of input. | Tests systematically unprompted, finds own bugs and states precise complexity. | Tests systematically and explains how they would test the code at scale or in production. |
Communication (20%)
| Level | 1 · Poor | 2 · Borderline | 3 · Solid | 4 · Outstanding |
|---|---|---|---|---|
| Intern / new grad | Stays silent for long stretches and cannot explain the code when asked. | Explains only when asked and struggles to say what the code does. | Thinks aloud when nudged, explains the plan and responds to hints. | Thinks aloud unprompted, explains each step and uses hints well. |
| Junior | Stays silent or gives explanations that do not match the code. | Narrates what they are doing only when the interviewer asks. | Thinks aloud and explains the plan before coding; takes hints well. | Explains each decision clearly and asks the interviewer useful questions. |
| Mid-level | Stays silent or speaks incoherently about the solution. | Narrates what they are doing only when the interviewer asks. | Thinks aloud and explains the reasons behind each decision. | Drives the conversation and adapts to hints gracefully without losing the thread. |
| Senior | Stays silent or speaks incoherently about the solution. | Explains decisions but needs the interviewer to steer the pacing. | Drives the session, explains decisions and their reasons, and manages time. | Drives the session, checks in at key points and adapts quickly to hints and pushback. |
| Staff | Stays silent or speaks incoherently about the solution. | Explains when asked but does not lead the discussion. | Leads like a design review: frames options, explains choices and manages time. | Leads crisply, invites challenge and tailors depth to the interviewer's questions. |
Company-specific signals
Get scored against this rubric
Thita's AI interviewer runs a live coding round and scores you against exactly this rubric, dimension by dimension, for the level you choose.
Practise a coding roundWhere this rubric comes from
These rubrics are researched: they follow the scoring methods big-tech companies describe publicly (behaviourally anchored ratings, bar raisers, leadership principles, named interview signals) and how their rounds are commonly run. No company publishes its actual scoring sheet, so the level-by-level wording is our own synthesis. It is the same rubric our AI interviewer scores every mock against.
Questions, answered
- How is a coding interview scored?
- A coding interview is scored on 5 dimensions: problem understanding (15%), approach and optimisation (25%), code correctness and quality (25%), testing and verification (15%), communication (20%). Each is rated 1 to 4 against anchors that rise with the level. At mid level, an outstanding score on approach and optimisation means: "Compares two or more approaches and chooses one with a stated reason."
- What is the difference between a mid-level and a senior coding answer?
- On approach and optimisation, the heaviest dimension, an outstanding mid-level answer "compares two or more approaches and chooses one with a stated reason", while an outstanding senior answer "reaches optimal unaided and explains when each alternative approach would be the better choice". The same answer can score 4 at mid level and 2 at senior: senior candidates are expected to raise things unprompted that mid-level candidates may be asked about.
- What does a score of 3 mean?
- Scores are relative to the level you are interviewing for. A 3 means solid for that level: the bar is met without help on the key points. A 4 means the candidate also raised the important points unprompted. A 2 usually means the interviewer had to lead the candidate to them.
- Is this the official rubric used by Google, Meta or Amazon?
- These rubrics are researched: they follow the scoring methods big-tech companies describe publicly (behaviourally anchored ratings, bar raisers, leadership principles, named interview signals) and how their rounds are commonly run. No company publishes its actual scoring sheet, so the level-by-level wording is our own synthesis. It is the same rubric our AI interviewer scores every mock against.