Lightning Talks + Closing Remarks DAY 1
summary
Five‑minute lightning set yields concrete wins and failures: converting PDFs to markdown with images using Gemini and GPT‑4 works for structure but returns wrong bounding boxes so image extraction with pimuPDF plus GP4 reattachment is the reliable hack; time zone canary detects DST duplicates and narrows plausible zones, saving manual sleuthing; a Streamlit clinical dashboard on AWS cuts deployment from months to days; a numpy structured‑dtype demo claims “maybe no need pandas.” Watch the demos showing code, data errors, and what actually runs.
AI-written summary · about
chapters (11)
- 0:00Lightning talks intro
- 3:40Convert docs to markdown with images
- 8:41Millionaire by data
- 16:27Files lying about time zones
- 21:13Readable ML test data
- 26:28Monitoring clinical trials
- 32:50NLP survival kit
- 36:47Missing data dangers
- 42:31Kaggle fellowship
- 48:44No need for pandas
- 53:37Closing remarks
