How TechInView scores an AI mock interview
When a round ends, the scoring model reads the full transcript of what you said to the interviewer and the code you wrote in the editor. It scores five dimensions from 0 to 100. Your overall score is the weighted sum of those five, and it maps to a hire recommendation from Strong Hire to No Hire.
This page covers DSA coding rounds. Other round types use their own rubrics, noted below.
The five dimensions
Problem solving and code carry the most weight, as they do in most big-tech coding loops. Communication, technical depth and testing make up the rest.
- 0130% weight
Problem Solving
Clarification, approach, edge cases
Did you ask about constraints and edge cases (empty input, duplicates, bounds) before writing code? Did you pick a sensible approach and explain why it fits? High scores go to candidates who change course when the interviewer points out a flaw instead of defending a dead end.
- 0225% weight
Code Quality
Readability, naming, idioms
The scorer reads your final code in the session editor: naming, structure, control flow and whether you added complexity you did not need. Ask yourself if a teammate could review it in two minutes. Small cleanups you make along the way count.
- 0320% weight
Communication
Thinking aloud, structured explanation
The interviewer hears how you explain the plan, talk through tradeoffs and respond to hints. You do not need a polished speech. Saying "I'm stuck on the duplicate case" scores better than two minutes of silence.
- 0415% weight
Technical Knowledge
Complexity analysis, trade-offs
Correct time and space complexity, a reason for each data structure you chose, and a fair comparison with the alternatives (extra memory vs. in-place, sort vs. hash map). Vague answers to follow-up questions pull this score down.
- 0510% weight
Testing
Edge cases, proactive testing
Walk an example through your code, name the edge cases and check them, either by running tests in the editor or by tracing by hand. Finding and fixing your own bug scores better than code that only handles the happy path.
- Overall score
The formula
Each dimension is scored 0 to 100, then weighted.
overall = Problem Solving × 30% + Code Quality × 25% + Communication × 20% + Technical Knowledge × 15% + Testing × 10%
Hire recommendation bands
The recommendation comes from the weighted overall score, not from any single dimension. A 95 in communication will not carry a 40 in problem solving.
| Recommendation | Overall |
|---|---|
| Strong Hire | 85–100 |
| Hire | 70–84 |
| Lean Hire | 55–69 |
| Lean No Hire | 40–54 |
| No Hire | 0–39 |
What you get after a round
A static example of the results page after a coding round with Tia: summary, radar chart and a note per dimension. Your real results page also has the transcript and your code.
Sample only · scores and notes are made up to show the format
Recommendation
Hire
Solid performance: clear approach, working solution, and reasonable complexity discussion. Communication was good with room to be more vocal during debugging. Overall aligned with a hire-level bar for this problem.
Your report is based on your own session and includes the transcript and your code.
Dimension breakdown
- 01 · 30% weight78/100
Problem Solving
You asked good clarifying questions about duplicates and empty inputs before coding. The two-pointer approach was appropriate; consider stating the invariant you maintain across moves earlier in the discussion.
- 02 · 25% weight82/100
Code Quality
Naming was clear and the loop structure was easy to follow. Minor nit: extracting the swap into a small helper would match common style for readability in longer solutions.
- 03 · 20% weight71/100
Communication
You explained your thinking at a steady pace. A few pauses were long; briefly narrating what you are stuck on helps Tia coach you faster.
- 04 · 15% weight74/100
Technical Knowledge
Time and space complexity were correct. You mentioned stability trade-offs when relevant; deepening one sentence on why the hash map beats sorting for this constraint would strengthen the answer.
- 05 · 10% weight68/100
Testing
You walked the main example and one edge case. Adding a quick check for single-element or all-equal inputs would mirror what many interviewers expect before they say “looks good.”
Technical Q&A, Engineering Manager and Behavioral
Technical Q&A, Engineering Manager and Behavioral rounds use a round rubric instead: problem solving, communication, technical depth, execution and judgment. Behavioral and Engineering Manager rounds also get a competency report graded against the value lens you choose at setup, such as Amazon Leadership Principles.
Want a scorecard of your own? Practice free, or start with a 5-minute AI interview.