Ai mock interview coach

Inspiration

Technical interviews are not just about knowing answers — they are about explaining thoughts clearly under pressure. Most candidates fail not because they lack knowledge, but because they have never practiced in a realistic interview environment. Existing mock interview platforms either: Overuse AI (making them expensive and unreliable), or Provide generic text-based practice with no real interview feel. We wanted to build something practical, fast, and honest — a system that simulates how real interviews actually happen, especially for technical roles.

What it does

AI Mock Interview Coach simulates a realistic technical interview experience using voice interaction and structured questioning. Users select a tech role (Frontend, Backend, Full Stack, etc.) The system conducts a 15-question interview using predefined, role-specific questions Users answer using voice or text After the interview, AI evaluates all answers in one pass A detailed dashboard shows: Overall score Strengths and weaknesses Clear improvement suggestions Role readiness assessment The focus is on real practice, not AI theatrics.

How we built it

The system is built with a cost-efficient and reliable architecture: Frontend: Next.js + React + TypeScript + Tailwind CSS Voice Layer: Web Speech API (speech-to-text & text-to-speech) Interview Logic: All interview questions are predefined and stored No AI is used during the interview itself Evaluation: A single AI call at the end evaluates all responses together Powered by Groq (LLaMA 3.3 70B) for fast, low-latency inference Result Dashboard: Clean UI with structured feedback and insights This architecture ensures stability, low cost, and a smooth demo experience.

Challenges we ran into

AI quota limitations: Free-tier AI APIs exhausted quickly during testing Voice reliability: Handling browser differences in speech recognition Balancing realism vs cost: Avoiding unnecessary AI calls without reducing quality Evaluation quality: Designing prompts that give honest, structured feedback instead of generic praise These challenges forced us to rethink the architecture and remove AI where it wasn’t truly needed.

Accomplishments that we're proud of

Built a fully functional mock interview system without AI dependency during interviews Reduced AI usage to one single call per interview Delivered realistic, role-specific interview flows Created a demo-safe, hackathon-ready product Designed a clean, professional UI that feels like a real interview tool

What we learned

AI should be used strategically, not everywhere Real-world constraints (cost, quotas, reliability) matter more than fancy features Good UX and system design can outperform brute-force AI usage Interview preparation is about communication, not just answers This project reinforced the importance of engineering discipline over hype.

What's next for Ai mock interview coach

Interview history and progress tracking Difficulty levels (Junior / Mid / Senior) Custom question sets per company or role Multi-language voice support Performance analytics over time Mobile app version Team and placement-cell integrations

Built With

• ai
• api
• css
• eslint
• groq
• javascript
• lucide
• next.js
• postcss
• radix
• react
• rest
• sdk
• speech
• tailwind
• typescript
• ui
• vercel
• web