How to get smarter while doing your job
summary
Shows concrete experiments: reproducing environments with Docker/Helm/Terraform, using console.log and the debugger as microscopes, and treating telemetry as instruments; finds 6 million useless session DB inserts and cuts them to stop an imminent RDS crash, runs a Lambda Graviton A/B where P99s are slower so feature flags roll it back to 1% then iterate until cost wins, runs a navbar A/B with no statistical significance yet, and uses ensemble programming and retrospectives to spread knowledge.
AI-written summary · about
chapters (11)
- 0:00Smarter software development
- 1:06Knowledge stock vs flow
- 4:17Four levels of experimenty work
- 10:46Experiments on your computer
- 15:26Testing distributed systems in production
- 21:32Measuring production impact
- 26:58Randomized control trials
- 31:56Action research for teams
- 40:06Ensemble programming experiments
- 45:42Scaling experiments
- 47:47Sharing experiments
related talks
41mDebug like a scientist
53mNDC Sydney Lightning Talks
33mGet Your Head Outta the Cloud – Let’s Talk Tech
43mLessons Learned Live Coding in Public
26mHacks to Work More Efficiently- Rain Leander
34mLightning Talks
47mDoing I.T. for SCIENCE! - Sprints, Startups and the Scientific Method
28mDev Tools to increase your speed that DON'T use AI