What was built
- Parses Python source into an Abstract Syntax Tree to understand structure and logic, and visualises the AST.
- Detects syntax errors and likely runtime issues with static analysis, and checks style against PEP 8.
- Uses Qwen2.5-7B-Instruct (via Hugging Face) to propose refactors and bug fixes, plus a chat interface for questions about the code.
- Reduced linting violations from 50+ to 35 per 1,000 lines of code (~30% improvement).
- Streamlit review platform delivers AI-generated feedback in under 2 seconds; built during the Infosys Springboard virtual internship.