How subashs27/ai-interview-management-system works
This is a Streamlit AI interview system. Candidates submit a resume, receive a profile-informed interview, answer questions by voice, and see an evaluation and report. The main stages are resume parsing and analysis, knowledge lookup or generation, adaptive question selection, transcription and answer evaluation, follow-up/completion decisions, and report display. An admin page and SQLite-backed knowledge and interview records are additional capabilities. The README documents the end-to-end flow; some internal links rely on source-index references rather than complete excerpts.
Main entry points: Candidate_v2.py Candidate interface · interview_manager.py Interview lifecycle · generator.py Knowledge generation · evaluation_engine.py Answer evaluation
Components 23
Candidate experience
- Candidate interface
pages/Candidate_v2.py
Profile and knowledge
- Resume parser
resume/resume_parser.py - Resume analyzer
resume/resume_analyzer.py - Knowledge retrieval
knowledge/knowledge_retriever.py - Knowledge generation
generator.py - Generation prompts
prompt.py
Interview runtime
- Interview lifecycle
interview/interview_manager.py - Question selection
interview/interview_engine.py - Adaptive decisions
interview/adaptive_engine.py - Interview state
interview/interview_state.py - Session access
interview/session.py - Voice recording
voice/recorder.py - Speech transcription
voice/speech_to_text.py - Voice analysis
voice/voice_analysis.py - Answer evaluation
evaluation/evaluation_engine.py - Evaluation prompt
evaluation/evaluation_prompt.py - Follow-up logic
interview/followup_engine.py - Interview report
interview/interview_manager.py
Admin and persistence
- Admin interface
pages/admin.py - SQLite records
database/database.py - Database schema
database/schema.py
Other components
- Candidate
- Gemini API
Connections 27
- Candidate → Candidate interface: uses
README.md - Candidate interface → Resume parser: uploads resume
README.md - Resume parser → Resume analyzer: provides text
README.md - Resume analyzer → Gemini API: analyzes with
resume/resume_analyzer.py - Candidate interface → Knowledge retrieval: retrieves knowledge
pages/Candidate_v2.py - Knowledge retrieval → SQLite records: looks up topics
knowledge/knowledge_retriever.py - Admin interface → Knowledge generation: generates topics
pages/admin.py - Knowledge generation → Generation prompts: uses prompt
generator.py - Knowledge generation → Gemini API: generates content
generator.py - Knowledge generation → SQLite records: saves knowledge
generator.py - Candidate interface → Interview lifecycle: starts interview
pages/Candidate_v2.py - Interview lifecycle → Session access: gets state
interview/interview_manager.py - Session access → Interview state: provides state
interview/session.py - Interview lifecycle → Question selection: selects questions
interview/interview_manager.py - Interview lifecycle → Adaptive decisions: chooses next step
interview/interview_manager.py - Candidate interface → Voice recording: records answer
pages/Candidate_v2.py - Candidate interface → Speech transcription: transcribes audio
pages/Candidate_v2.py - Speech transcription → Gemini API: transcribes with
voice/speech_to_text.py - Candidate interface → Voice analysis: analyzes voice
pages/Candidate_v2.py - Interview lifecycle → Answer evaluation: evaluates answers
interview/interview_manager.py - Answer evaluation → Evaluation prompt: uses prompt
evaluation/evaluation_engine.py - Answer evaluation → Gemini API: evaluates with
evaluation/evaluation_engine.py - Interview lifecycle → Follow-up logic: handles follow-ups
- Interview lifecycle → Interview report: builds report
interview/interview_manager.py - Candidate interface → Interview report: displays report
pages/Candidate_v2.py - Admin interface → SQLite records: reads records
- Candidate interface → Database schema: creates tables
pages/Candidate_v2.py