AI Mock Interview System with Feedback

The AI Mock Interview System with Feedback is a final-year project that plays the role of a real interviewer: it asks technical and HR questions, listens to your spoken answers, transcribes them with speech recognition, and scores each answer on content accuracy, confidence and communication using a large language model. InterviewAI — the Streamlit app at its heart — then shows a dashboard with per-question scores, concrete improvement tips and an overall readiness meter, so students walk into placement season prepared instead of anxious.

Campus placements are where theory meets pressure: students who know the answers still freeze, ramble or miss the point under interview stress. A system that lets them rehearse with instant, honest, rubric-based feedback — any number of times, at zero cost — is both technically rich and instantly demoable in a viva.

This is a built-to-order project: the question banks, scoring rubric and feedback prompts are configured for your build, and the full pipeline ships with documentation you can defend question by question.

1. The student picks a round (technical or HR), a topic and a difficulty level.
2. InterviewAI asks a question — displayed on screen and optionally read aloud.
3. The student answers by speaking; Whisper transcribes the audio to text in real time.
4. The transcript, question and rubric are sent to the LLM, which scores the answer on content, confidence and communication and writes specific feedback.
5. The dashboard updates: per-question scores, improvement tips, a model answer for comparison, and the overall readiness meter.

• AI-conducted mock interviews — technical (DSA, DBMS, OS, CN, projects) and HR rounds
• Speech-to-text pipeline (Whisper / speech recognition) transcribes spoken answers in real time
• LLM-based scoring rubric: content accuracy, depth, confidence, clarity and communication
• Per-question scorecards with strengths, weaknesses and model-answer comparisons
• Personalized improvement tips generated after every session
• Readiness meter — an aggregate score tracking progress across sessions
• Question bank with difficulty levels and topic tags; add your own questions
• Session history and progress charts in the Streamlit dashboard
• Text-input mode for quiet environments (no microphone needed)

• Full Python source code (Streamlit app + pipeline), well-commented
• Configured question bank (technical + HR, multiple difficulty levels)
• LLM scoring-rubric prompt templates (tunable)
• Project report PDF (problem statement, architecture, pipeline, results, conclusion)
• PPT presentation for final review
• Viva Q&A preparation document (LLM evaluation, STT pipeline, rubric design questions)
• Setup guide (API keys, microphone configuration, running the app)

FAQs

1. Is this AI interview project suitable for a final-year CSE/IT review? Yes — it combines speech processing, prompt engineering, LLM-based evaluation and a full-stack dashboard. Examiners can probe the STT pipeline, rubric design, and how you handle LLM hallucinations, and the viva Q&A covers all of it.
2. Do I need to pay for the LLM API? The system needs an API key, but a free-tier key or a small pay-as-you-go balance is plenty for development, demos and your review. The setup guide walks through key configuration.
3. How does the scoring actually work? Each answer is evaluated against a structured rubric prompt — content accuracy, depth, confidence, clarity, communication, each scored 1–10 — and the LLM returns scores plus written justification and tips. You can show examiners the exact prompt templates.
4. What if there's no microphone during the demo? Text-input mode lets you type answers instead of speaking — the same scoring and feedback pipeline runs on typed text. Many students demo this way in noisy review halls.
5. Can I add my own questions? Yes — the SQLite question bank is fully editable. Add company-specific questions (TCS NQT, Infosys, product companies) with topic tags and difficulty levels, which is a strong personalization story for your review.
6. Does it really help with placements? That's the pitch: unlimited, judgment-free mock interviews with instant rubric-based feedback and a readiness meter that tracks improvement across sessions. It's the rare final-year project your juniors will actually want to use.