AI interview practice with real-time feedback on what to fix.
The Scoring Engine

Every score, explained.

Most AI scoring is a black box. Ours isn't. Scroll through the exact pipeline your session runs through — from voice answer to final score. No vibes, just math.

0
Signals tracked
0
Scoring layers
0
Black-box magic
Live audio signalanalysing turn 4
Scroll to follow the pipeline
01 / 09Ingest

Your voice answer enters the engine.

The moment you stop speaking, audio + transcript are captured and queued for analysis.

Audio · 00:00 → 00:42turn 4 of 8
96kHz · 16-bit42.3s
Live transcript

“So when we were building the inventory sync, the main bug was that our cache wasn't invalidating when two warehouses pushed updates at the same time. We had to add a versioning layer…”

inventory synccache invalidationversioning
02 / 09Parse

Transcript + voice signals split apart.

The audio gives us pace, pauses, filler words. The transcript gives us substance. Both flow forward in parallel.

Voice signals
Pace142 wpm
Filler words7 detected
Pause ratio8%
Confidence (tonal)72%
Content signals
Word count94 words
Key anchors hit3 of 4
Topic driftOn-topic
Example usedYes (project)
03 / 09Classify

Every turn gets a status tag.

An LLM tags how strong each turn was. The tag drives the math downstream.

Turn classifier output
CONFIDENT
0%

Clear, specific, with example

NEEDS_EVIDENCE
0%

On track, lacks specifics

HAND_WAVY
0%

Vague, abstract, no anchor

DONT_KNOW
0%

Stalled or admitted blank

04 / 09Layer 1 · The 14 signals

14 signals get measured for every question.

Grouped into three buckets. Each signal is a small, transparent calculation — not a vibe.

Answer Quality4 signals
  • Relevance0
    OFF_TOPIC 10% / PARTIAL 60% / ANSWERED 100%
  • Completeness0
    Anchors hit vs total anchors expected
  • Specificity0
    Per-turn specificity score, averaged
  • Depth0
    CONFIDENT/NEEDS_EVIDENCE/HAND_WAVY/DONT_KNOW
Communication4 signals
  • Clarity0
    Penalizes vague & hand-wavy turns
  • Conciseness0
    Bell curve · optimal 30–100 words/turn
  • Coherence0
    Per-turn topic-drift penalty
  • Confidence0
    AI confidence score per turn, averaged
Behavior6 signals
  • Hesitation0
    True if turn-1 was DONT_KNOW
  • Follow-up responsiveness0
    Specificity gain turn 1 → final turn
  • Recovery after hint0
    Status after RESCUE_HINT
  • Self-correction0
    Specificity jump ≥ threshold
  • Composure0
    Penalizes excess time, retries, hesitation
  • Example usage0
    True if any turn = CONFIDENT
05 / 09Per-question score

The 14 signals collapse into one question score.

A single number for that question — the building block of everything that follows.

Core drivers
Relevance100
Completeness85
Specificity78
Depth60
Q score0/100
avg of 4 core drivers
06 / 09Layer 2 · Section roll-up

Question scores combine into section scores.

Questions group by interview phase — Technical, Behavioral, Wrap-up. Each section gets its own score.

Intro & WarmupSoft phase
0
Q1
88
Q2
91
Technical / Deep Dive
0
Q1
81
Q2
74
Q3
69
Q4
78
Behavioral / Situational
0
Q1
82
Q2
87
Wrap-upSoft phase
0
Q1
93
07 / 09Layer 3 · Weighting

Mode-specific weights are applied.

A Job Interview weights technical depth differently than an HR round. The same section score lands at a different final number depending on what you're prepping for.

Job Interview
Core phases1.5×
Soft phases0.2×
HR Round
Core phases1.2×
Soft phases1.3×
Personal Interview
Core phases1.2×
Soft phases0.4×
Core phases: Technical · Deep Dive · Exploration · Situational·Soft phases: Intro · Warmup · Wrap-up
08 / 09Protection gates

Three evidence gates run before the final score.

If there isn't enough signal to score you fairly, we cap the score and tell you why. No phantom 95% scores from 30-second sessions.

Insufficient data
< 2 substantive answers
Caps score at
3%

Not enough turns to score anyone fairly. We tell you to do a longer session.

Mostly missed
> 50% off-topic or unanswered
Caps score at
25%

Majority of the session had no useful content. A high score would be misleading.

Narrow scope
Only 1 competency tested
Caps score at
40%

Acing one area doesn't prove broad readiness. We weight it accordingly.

09 / 09Final result

One score. With trend and difficulty context.

Everything above rolls into a single number — but the number alone isn't the story. Trend and difficulty tell you whether you're actually improving.

Final0/100
Session complete
Strong session. Technical clarity held up under pressure. Specificity dropped on the system-design follow-ups — bring data to back claims next time.
Session trend
+14pointsImproving
Session 1Now
First-half avg 68 → second-half avg 82
Difficulty reached
Easy
Medium
Hard
Hard unlocks once Medium-tier accuracy averages above 80%.
That’s the whole rubric

Try a session.
See your number — and the math behind it.

Every report you get points back to the signals above. No surprises.