AI Interview Intelligence System

A full-stack AI Interview Coach that runs locally with Flask, SQLite, Groq cloud API or Ollama (LLM), and OpenAI Whisper (speech-to-text). Candidates register, upload a resume, choose a role and company, answer role-specific questions (text or voice), receive instant multi-dimension feedback, and may rewrite each answer once for a re-score.

Features

• Candidate registration with resume parsing (PDF/DOCX) and skill extraction
• Multi-round interviews tailored to companies (Amazon, Google, Microsoft, Meta, etc.)
• Role-specific questions for SWE, PM, DS, QA, DevOps, and more
• LLM (Groq cloud API or local Ollama) for question generation and answer evaluation
• Local speech-to-text (Whisper) — record or upload audio, transcript is auto-filled
• Filler-word detection (um, uh, like, you know, etc.) with score impact
• One rewrite per answer — see feedback, rewrite, get a second evaluation
• Persistent history in SQLite — sessions, answers, skill gaps, recommendations
• Dashboard with progress charts, skill radar, and per-session comparison

Requirements

• Python 3.10+
• ffmpeg on PATH (required by Whisper)
• Either a Groq API key (cloud, recommended for deployment), or Ollama running locally (ollama serve) for local-only use

Install ffmpeg on Windows

# Using winget (recommended)
winget install Gyan.FFmpeg

# Or download from https://www.gyan.dev/ffmpeg/builds/ and add bin/ to PATH
ffmpeg -version  # verify

Set up LLM provider

Option A — Groq (cloud, preferred for Render deployment):

1. Get a free API key from https://console.groq.com/keys
2. Copy .env.example to .env and set your key:

   GROQ_API_KEY=gsk_your_key_here
   

Option B — Ollama (local, easy for development):

ollama serve               # in one terminal
ollama pull llama3.2:latest

Install Python dependencies

cd ai-interview-system
python -m pip install -r requirements.txt

Whisper downloads the base model on first transcription (~140 MB). Override with the environment variable WHISPER_MODEL=tiny|base|small|medium|large.

Environment Variables

Create a .env file (copy from .env.example) with these optional overrides:

Variable	Default	Description
GROQ_API_KEY	—	Groq API key (set this on Render; omit for local Ollama)
GROQ_MODEL	openai/gpt-oss-120b	Groq model ID
OLLAMA_BASE_URL	http://localhost:11434	Ollama server URL (fallback when no GROQ_API_KEY)
OLLAMA_MODEL	llama3.2:latest	Ollama model name
WHISPER_MODEL	base	Whisper model size: tiny, base, small, medium, large
SECRET_KEY	random	Flask session secret (set a fixed value on Render to persist sessions across restarts)

Deploy to Render

This app requires ffmpeg at the system level (the Python package ffmpeg-python is only a wrapper — it does not bundle the binary). Render's default Python runtime does not include ffmpeg, so you need a render.yaml with a pre-build command or a Dockerfile.

Important: On Render you must set GROQ_API_KEY in the environment variables (Groq is a cloud API that works from any server). Ollama only runs locally and will not be accessible from Render.

Option A — render.yaml (Blueprint)

services:
  - type: web
    name: ai-interview-system
    env: python
    buildCommand: |
      apt-get update && apt-get install -y ffmpeg
      pip install -r requirements.txt
    startCommand: gunicorn app:app --bind 0.0.0.0:$PORT --workers 1 --timeout 300
    envVars:
      - key: GROQ_API_KEY
        sync: false          # Enter manually in Render dashboard (never committed)
      - key: GROQ_MODEL
        value: openai/gpt-oss-120b
      - key: WHISPER_MODEL
        value: tiny
      - key: SECRET_KEY
        sync: false
      - key: PYTHON_VERSION
        value: 3.11.9

Option B — Dockerfile

FROM python:3.11-slim
RUN apt-get update && apt-get install -y ffmpeg && rm -rf /var/lib/apt/lists/*
WORKDIR /app
COPY requirements.txt .
RUN pip install --no-cache-dir -r requirements.txt
COPY . .
CMD gunicorn app:app --bind 0.0.0.0:$PORT --workers 1 --timeout 300

Set GROQ_API_KEY in Render's dashboard (Environment Variables → Add Secret File or manual entry). Use WHISPER_MODEL=tiny to reduce cold-start download time. Persistent storage (SQLite) will reset on each deploy — mount a Render Disk or switch to PostgreSQL for production.

Run

# Local dev with Ollama (no API key needed):
python app.py

# Or with Groq (set .env first):
# GROQ_API_KEY=gsk_... python app.py

# Server: http://127.0.0.1:5050

Run tests

python -m pytest tests/ -v

Project layout

ai-interview-system/
├── app.py                 # Flask routes
├── ai_service.py          # LLM (Groq / Ollama) prompts + role rubrics
├── interview_engine.py    # Session state, rounds, evaluation
├── stt_service.py         # Whisper STT + filler-word detection
├── resume_parser.py       # PDF/DOCX resume parsing
├── company_rounds.py      # Per-company round structures
├── coding_questions_bank.py
├── aptitude_bank.py       # Aptitude MCQ bank
├── templates/             # HTML pages
├── static/                # CSS/JS
└── tests/                 # pytest suite