Day 3
summary
Day 3 packs practical takeaways: an insurance anomaly system uses unsupervised ML (robust PCA/autoencoder style), spends ~70% effort on feature engineering, reaches 84% of outliers detected and ships in R after 4–5 months; climate tech analysis pins 73% of emissions to energy and pushes solar + cheap storage (battery cost math: today ~\$70T to store a day, target ~$20/kWh), with nuclear discussed as baseline; VR/AR work stresses hand‑tracking and physics vs kinematic UX (Unity prototyping, holistic tests); Kotlin Multiplatform warns about strict native memory, coroutine limits and shows ~80% shared‑code wins after rework; Quake postmortem reveals brush editor tradeoffs, Pentium FDiv bug and brutal crunch.
AI-written summary · about
chapters (8)
- 0:00Welcome to day three of WeAreDevelopers Live Week!
- 8:50Anomaly Detection – Using unsupervised Machine Learning for detecting anomalies in customer base – Lukas Kölbl
- 1:36:04How will we live and work after COVID-19? – Panel discussion with Heike Mensi-Klarbach, Christine Catasta, Martin Wezowski, Oliver Blüher, moderated by Benjamin Ruschin
- 3:06:53Can technology solve climate change? – Eric Steinberger
- 4:20:47Evolving from pancake to spatial computing: Creating a VR/AR ready developer mindset – Roger Kung & Dennys Kuhnert
- 5:04:56Kotlin Multiplatform – true power of native code reuse – Petar Marijanović
- 6:21:30The Quake Postmortem – John Romero
- 7:36:00Hidden histories of women in code – Brenda Romero
