pintarkristian/ai-powered-code-bug-classifier

Source Code Bug Classifier is an end-to-end ML/AI repository that predicts whether a source-code snippet is clean (0) or potentially buggy/risky (1). It compares two modeling tracks: a lightweight TensorFlow/Keras baseline on tokenized code and a PyTorch Transformer (CodeBERT-style) fine-tuned for binary classification. The project includes Pandas-based preprocessing, static feature extraction, evaluation reports, a FastAPI inference API, Dockerization, automated tests, and CI. Who uses it: ML/AI engineers, data scientists, MLOps practitioners, and developers building code-quality or defect-detection tooling or demonstrating skills in a portfolio. Key features: - Dual modeling tracks: TensorFlow baseline and PyTorch Transformer fine-tuning - Data pipeline and static code features (Pandas) - Metrics, evaluation reports, and visualization - FastAPI endpoint for inference, Docker-ready deployment - Tests (pytest), CI (GitHub Actions), and reproducible project layout with scripts and notebooks

このRepoのdescription / GitHub topics / READMEのAI要約に出現したAI関連キーワードをカテゴリ別に表示。各バッジは該当カテゴリの詳細ランキングへリンクします。