How elangkumaran2004/ai-mock-interviewer works
AI Mock Interviewer is a local Streamlit app for timed technical or HR interview practice. Candidates choose topics, level, and duration, then answer each generated question by typing or speaking; answers are scored and the completed session produces a coaching report and Markdown download. The graph follows the documented workflow and sampled modules. Some UI calls are in omitted app.py excerpts, so those relationships are grounded in the README rather than presented as inspected source details.
Main entry points: app.py Streamlit screens · questions.py Question writer · judge.py Answer judge · llm.py Structured LLM client
Components 12
Interview experience
- Streamlit screens
app.py - Interview session
interviewer/session.py - Voice transcription
interviewer/speech.py
Question generation
- Question writer
interviewer/questions.py - Topics and levels
interviewer/config.py
Answer evaluation
- Answer judge
interviewer/judge.py - Report builder
interviewer/report.py
Shared services
- Validated data models
interviewer/schemas.py - Structured LLM client
interviewer/llm.py
Other components
- Candidate
- Ollama model
- faster-whisper
Connections 17
- Candidate → Streamlit screens: sets up and answers
README.md - Streamlit screens → Question writer: requests first question
README.md - Streamlit screens → Interview session: starts and advances
README.md - Streamlit screens → Voice transcription: submits recording
README.md - Voice transcription → faster-whisper: transcribes audio
README.md - Streamlit screens → Answer judge: submits answer
README.md - Interview session → Question writer: writes next question
README.md - Question writer → Topics and levels: uses topics and levels
interviewer/questions.py - Question writer → Structured LLM client: generates questions
interviewer/questions.py - Answer judge → Topics and levels: uses level rubric
interviewer/judge.py - Answer judge → Structured LLM client: scores answer
interviewer/judge.py - Report builder → Structured LLM client: writes coaching summary
interviewer/report.py - Report builder → Streamlit screens: provides report and download
README.md - Structured LLM client → Ollama model: requests structured output
interviewer/llm.py - Question writer → Validated data models: validates generated data
interviewer/questions.py - Answer judge → Validated data models: returns score model
interviewer/judge.py - Report builder → Validated data models: returns summary model
interviewer/report.py