Leveraged Python and n8n to orchestrate workflows between Google Drive (document ingestion), Gemini (data extraction), and Telegram (notifications).
Used AI coding assistants (Replit, Claude, Codex) extensively for debugging, refactoring, and workflow optimization throughout the development lifecycle.
Optimized video/audio handling and merging via yt-dlp and FFmpeg integrations.
Engineered dynamic scikit-learn Pipelines with automated preprocessors (ColumnTransformer, StandardScaler, OneHotEncoder) supporting both classification and regression tasks dynamically based on user input.
Integrated GridSearchCV for automated hyperparameter tuning and optional SMOTE oversampling to combat imbalanced classification datasets natively within the pipeline.
Handled live video capture streams via cv2.VideoCapture.