Interview Review

Local-first interview review workspace built with Streamlit.

The app takes a recorded interview, generates a transcript with speaker attribution, extracts high-value QA pairs, produces AI reference answers for each question, and renders a readable review report. It also supports resume import, single-run history, cross-interview comparison, and a global capability profile.

Note: This tool is primarily designed for Chinese-language interviews. The ASR processing, QA extraction, and reporting pipelines are optimized for Mandarin Chinese content.

Features

• Local Windows workflow for large interview audio and video files
• Tencent Cloud ASR transcription with COS upload for large media
• High-value QA extraction from full interview transcript
• AI reference answers and per-question review
• Markdown review report generation
• Resume import from PDF / DOCX / TXT / MD
• History, growth comparison, multi-run trend view, dashboard, and profile pages

Tech Stack

• Python
• Streamlit
• Tencent Cloud ASR (Chinese ASR optimized)
• Tencent Cloud COS
• Kimi / Moonshot API

Requirements

• Windows
• Python 3.11 to 3.13 validated
• Python 3.14 may run, but some dependencies currently emit compatibility warnings
• FFmpeg and FFprobe available in PATH
• Tencent Cloud credentials for ASR and COS
• Kimi / Moonshot API key

Quick Start

1. Create a virtual environment and install dependencies:

python -m venv .venv
.\.venv\Scripts\python.exe -m pip install -r requirements.txt

2. Copy the example environment file:

copy .env.example .env

3. Fill in the required values in .env.

4. Run the built-in self-check:

.\.venv\Scripts\python.exe -m scripts.self_check

The self-check creates the local runtime workspace automatically, including runtime/uploads for browser-uploaded files.

5. Start the app:

.\.venv\Scripts\python.exe -m streamlit run app.py

Or use the Windows helper:

.\start_app.bat

Environment check only:

.\start_app.bat --check

Environment Variables

Required in normal use:

• TENCENT_SECRET_ID
• TENCENT_SECRET_KEY
• ASR_REGION
• COS_REGION
• COS_BUCKET
• LLM_API_KEY

Important optional values:

• ASR_ENGINE_MODEL
• COS_PREFIX
• COS_URL_EXPIRE_SECONDS
• LLM_BASE_URL
• LLM_MODEL
• LLM_STRUCTURED_MODEL
• FFMPEG_BINARY
• FFPROBE_BINARY
• RUNTIME_DIR
• UPLOAD_DIR

See .env.example for the full template.

Usage Flow

1. Open New Analysis
2. Choose a local interview media file
3. Run the full pipeline
4. Review:
   • transcript text
   • extracted QA pairs
   • AI reference answers
   • final report
5. Use History, Growth, and Profile for follow-up review, trend tracking, and cross-run comparison

Project Structure

• app.py: app entry
• pages/: Streamlit pages
• ui/: shared UI shell and analysis renderers
• part_a/: validation, audio extraction, COS, ASR
• part_b/: QA extraction, analysis, reporting
• services/: repository and orchestration services
• scripts/: self-check and gate runners
• runtime/: local upload workspace for browser-side files
• tests/: gate and runtime tests

Development Checks

.\.venv\Scripts\python.exe -m scripts.self_check
.\.venv\Scripts\python.exe -m scripts.run_phase1_gate
.\.venv\Scripts\python.exe -m scripts.run_phase2_gate
.\.venv\Scripts\python.exe -m scripts.run_phase3_gate
.\.venv\Scripts\python.exe -m scripts.run_phase4_gate
.\.venv\Scripts\python.exe -m scripts.run_phase5_gate
.\.venv\Scripts\python.exe -m scripts.run_phase6_gate

Security Notes

• Never commit .env
• Never commit real interview videos, transcripts, reports, or resumes
• Rotate any credential that was ever shared in chat, screenshots, or documents
• Use a low-privilege Tencent Cloud key for development

See SECURITY.md.

Contributing

See CONTRIBUTING.md.

License

This project is released under the MIT License. See LICENSE.

Limitations

• Language Support: Currently optimized for Chinese (Mandarin) interviews only. English and other language support is planned for future releases.
• Cloud Dependency: Requires Tencent Cloud ASR/COS services (China region). You may need to adapt the code for other cloud providers in your region.
• Platform: Primarily tested on Windows. Linux/macOS support is experimental.