• Claude Code: Used for hardware-specific metric optimization and system profiling guidance to maximize local compute resources.
• Gemini: Used to translate the initial Project_Details_Announcement.pdf into a structured roadmap, and for guidance on algorithm selection, architectural design (e.g., ADWIN + ARF), and diagnosing complex ML issues such as temporal data leakage and the curse of dimensionality.
• GitHub Copilot: Used exclusively for syntax autocomplete and generating brief commit messages. (Version control was restricted to the codebase only; sensitive data remained strictly local).
• Notion AI: Used strictly to format and structure my rough phase notes into a readable changelog. It was not used to generate the final analytical documentation.
• Google Search AI Overview: Acknowledged for transparency. Automatically provided high-level summaries while I researched specific bug-fix solutions (e.g., Dask KilledWorker errors) and alternative data ingestion methods.