LIGHTNING TALKS
summary
Lightning talks run five minutes each and hit concrete takeaways: GPU Python pathway advice with CuPy and RAPIDS and reusable CUDA kernels plus an Accelerated Computing Hub and free DLI course; a topic‑modeling pipeline using LLM embeddings that labels trends and finds 80% novel talks across 2016–2025; a 78‑card tarot experiment feeding GPT that produced actionable advice; meetup platform failures and migration strategies; Jupytext paired notebooks for script/output sync; pandas autocorrelation showing 12‑ and 6‑month peaks; PowerBI + scikit‑learn via pickled models; and the reminder to actually read errors.
AI-written summary · about
chapters (10)
- 0:00Lightning talks format
- 1:29Python on GPU with CUDA
- 7:09Talk novelty with LLMs
- 11:41AI tarot for customer service
- 18:33Meetup platform decline
- 24:59Jupytext for Jupyter
- 33:41Pandas autocorrelation
- 39:02Power BI with Python
- 42:15Read the error message
- 44:44Closing and Q&A








