Projects
AI Interviewer
AI Interviewer generates tailored interview questions and provides feedback to improve readiness. It emphasizes prompt design, response evaluation, and a feedback loop that encourages practice. The key challenge was creating a balanced evaluation system that feels helpful without being overly rigid.
The problem this solves
Practicing for technical and behavioral interviews is hard without a real interviewer. Most people repeat the same answers to the same questions without knowing whether they're actually improving. This project gives candidates a simulated interview experience with real-time AI feedback — so every practice session has a clear outcome.
What it does
- Role-specific question generation for technical and behavioral rounds
- AI-powered response evaluation with scoring and feedback
- Session recording to track improvement over time
- Adjustable difficulty and topic focus per session
- Clean interview UI designed to reduce anxiety, not add to it
- Detailed post-session summary with areas to improve
Hard parts and how I solved them
Getting AI feedback that feels genuinely useful — not generic — was the central challenge. Simply asking the model to 'evaluate this answer' produces vague, useless responses. I iterated extensively on prompt structure, breaking evaluation into specific dimensions: clarity, relevance, depth, and communication. The second challenge was preventing the question generator from becoming repetitive across sessions — which required building session context into the prompt so the model knew what had already been asked.
Tech stack
Outcome
A working mock interview tool that generates context-aware questions and delivers specific, actionable feedback per answer. The feedback loop is tight enough that users can see noticeable improvement across sessions if they engage with the suggestions.
My role
End-to-end design and development — AI prompt engineering, response evaluation logic, session state management, and the full React frontend.