Performance analytics, built in

InterviewIQ

A live AI interviewer that reads the answer behind your answer.

Experience interviews with an intelligent AI interviewer that speaks naturally, understands your voice, and evaluates every response through the MLIM pipeline—measuring emotion, intent, communication quality, and behavioral consistency to deliver deep, actionable insights.

4-layer intent model7 emotion classesLive voice & camera
Session report
Live session
82/ 100
Integrity score98%
Weak areas flagged3
Questions answered8 / 8
Score across sessions
Session 1Session 5
System design92%
Communication78%
Problem solving85%
Meet Your Interviewer

This isn't a form.
It's an interview.

An animated avatar runs the session end to end — reading each question aloud, waiting on your spoken or typed answer, and reacting in real time. Your camera and microphone stay active the whole way through.

Speaks every question aloud

A lip-synced animated avatar reads each question through your browser's speech synthesis, mouth and gaze moving in real time.

Listens to your real voice

Answer out loud. Groq Whisper Large v3 transcribes your recording live and feeds it straight into the evaluation pipeline.

Watches your expressions

Your webcam feed is read continuously across 7 emotion classes, fused alongside what you actually say.

Follows up when unsure

If your answer's intent is ambiguous, the interviewer asks a targeted clarification instead of moving on blind.

Speaking now

“Tell me about a time you had to push back on a technical decision.”

Two Ways To Train

Practice mode to learn.
Simulation mode to prove it.

Practice Mode

Learn as you go
Structured feedback card after every single answer
See your MLIM read on tone, clarity, and confidence immediately
Pause, reflect, and retry questions without pressure

Simulation Mode

Feel the real thing
Interviewer stays neutral — brief acknowledgements only
No hints, no mid-session scoring, just like a real panel
Full evaluation and report generated only once you finish
Running Underneath

The 4-LayerIntent Pipeline

Every spoken or typed answer in your interview passes through all four layers. ASL and PEL run first, then GSTL updates the HMM belief state, and IFL fuses everything into a final intent prediction with entropy-based uncertainty scoring.

ASL and PEL run on the fast 8B model for low latency
HMM belief state persisted across the full session
Entropy threshold automatically triggers a clarification question
IFL runs on the 70B reasoning model with full session history
01
ASLAffective Signal Layer

Valence-arousal extraction, sentiment polarity classification, emotional uncertainty estimation, and affective masking detection across voice transcripts and typed responses.

ValenceArousalMasking DetectionSentiment Polarity
02
PELPragmatic Encoding Layer
03
GSTLGoal-State Tracking Layer
04
IFLIntent Fusion Layer
IFL Output — Sample Intent Distribution
genuine_answer
64%
face_saving_assertion
14%
seeking_validation
11%
expressing_confusion
7%
committed_retry
4%
Capabilities

Every layer of intelligence,
running on your interview.

From facial expression to pragmatic speech acts — the full analysis stack activates live during your session.

Real-Time Affect Engine

Valence-arousal estimation, stress indicators, and affective-masking detection computed on every submitted answer.

Facial Expression Analysis

Live detection across 7 emotion classes via face-api.js, fused into the MLIM affective pipeline alongside your typed or spoken answer.

Goal-State Belief Tracker

An HMM belief distribution maintained across the full interview session, tracking goal drift and trajectory shifts turn by turn.

Session Integrity Monitor

Tab-switch, window-blur, copy-paste, right-click, and DevTools detection, with camera/mic suspension and an integrity score in your final report.

Whisper Voice Transcription

Groq Whisper Large v3 transcribes your recorded answer and feeds the identical text into the same MLIM pipeline as typed responses.

Entropy Clarification Gate

When IFL uncertainty crosses threshold, the interviewer automatically asks a targeted follow-up instead of moving on blind.

Process

From job description
to full assessment report.

01

Configure your session

Select your target role and paste the job description. The Groq-powered question engine generates fresh technical, behavioral, and scenario questions for that exact role.

02

AI interviewer takes over

An animated avatar reads each question aloud through your browser's speech synthesis, with mouth movement synced to speech. Camera and microphone activate for the session.

03

Answer by voice or text

Speak naturally into your mic or type your answer. Groq Whisper Large v3 transcribes voice in real time. Both inputs feed the identical MLIM pipeline.

04

MLIM pipeline processes

ASL and PEL run first. GSTL updates the HMM goal belief state. IFL fuses every signal with your session history into a final intent label.

05

Live analytics refresh

In Practice mode, a structured feedback card scores your answer. In Simulation mode, the interviewer stays neutral and evaluation runs after you finish.

06

Report generated

Your session closes with a full breakdown: overall score, weak areas, recommended topics, per-question detail, and an MLIM summary — exportable as PDF.

The Payoff

Walk away with a
graded interview report.

Every session closes with an overall score, an integrity score, per-question detail, flagged weak areas, recommended topics to revisit, and a full MLIM summary — exportable as a PDF.

82/100
Overall Score
98%
Integrity Score
3 flagged
Weak Areas
PDF ready
Export
0
MLIM Layers
ASL · PEL · GSTL · IFL
0
Intent Labels
Searle speech-act taxonomy
0
Emotion Axes
face-api.js detection
0
Groq Models
8B fast · 70B reasoning
FAQ

Common questions

Free · No credit card required

Ready to face your AI interviewer?

Create an account and start your first session. Watch the interviewer speak, listen, and read your intent across all four MLIM layers, live.