🎓 Data Interview Coach

AI-powered interview practice tool for Data Analyst / Analytics Engineer roles.
Practice SQL, behavioral, and project-deep questions — Claude evaluates each answer live with streaming feedback.

🤗 Live Demo on HuggingFace Spaces →
Seeded with 6 demo sessions showing a realistic 3-week improvement arc. CLI requires local setup with your own Anthropic API key.

---

Architecture

┌──────────────┐   answer    ┌─────────────────────┐   streaming   ┌───────────────────┐
│   👤 User    │──────────→  │   🖥️  coach.py       │─────────────→ │  🤖 Claude Sonnet  │
│              │ ←────────── │   Rich CLI           │ ←──────────── │  API + caching    │
└──────────────┘  feedback   └──────────┬──────────┘   scored eval  └───────────────────┘
                                        │ save session
                              ┌─────────▼─────────┐
                              │  💾 SQLite DB      │
                              │  interview.db      │
                              └─────────┬─────────┘
                                        │ read history
                              ┌─────────▼─────────┐      deploy     ┌────────────────┐
                              │  📊 dashboard.py   │───────────────→ │  🤗 HF Spaces  │
                              │  Streamlit         │                 │  Docker        │
                              └───────────────────┘                 └────────────────┘

---

Practice modes

Mode	Questions	Focus
SQL Drill	12	Window functions · CTEs · UNNEST · LAG · Cohort · Dedup · WHERE vs HAVING
Behavioral	9	STAR stories · Tell me about yourself · Weakness · 5-year goal
Project Deep	8	Olist · Uber · Weather Pipeline · MCP Agent · SO Survey · HR BI · Interview Coach
Stats & A/B Testing	6	Mean vs Median · p-value · Type I/II errors · A/B test design · Pitfalls
Mixed	12	All categories — real interview simulation

---

Dashboard

📊 KPI Overview

📈 Score Analytics

🎯 Skills Radar

📋 Performance by Topic

📅 Session History

📝 Recent Answers & Claude Feedback

Each answer is scored 1–10 with specific, actionable feedback from Claude Sonnet.

---

Features

• Streaming feedback — Claude's evaluation appears word-by-word in the terminal
• Prompt caching — system prompt cached across questions (~70% cost reduction)
• Progress tracking — SQLite stores every answer; dashboard shows trend, radar, weak spots
• Interview Readiness gauge — weighted score (SQL 40% · Behavioral 35% · Project 25%)
• Practice Next panel — auto-recommends the 3 weakest topics each session

---

Stack

Python 3.11 · Anthropic Claude API (streaming + prompt caching) · Rich · Streamlit · Plotly · SQLite · Docker

---

Local setup

git clone https://github.com/evgeniimatveev/interview-coach
cd interview-coach
python -m venv .venv && .venv\Scripts\activate   # Windows
pip install -r requirements.txt
cp .env.example .env                              # add ANTHROPIC_API_KEY
python coach.py

streamlit run dashboard.py   # dashboard only (no API key needed)

---

Built by Evgenii Matveev · May 2026