Making smart investments: A framework for maximizing your ROI in technical decisions
summary
Built a state‑of‑the‑art machine learning model, spent about half a year coordinating with a features team, ran AB tests and finds little to no impact, so the model never justifies rollout; evaluated replacing a core platform, saw request load jump from ~100,000 to millions per second, rejects vendor, scales in‑house at the cost of much higher cloud spend and temporary burnout; offers a repeatable ROI checklist using measured potential, quantified risks, clear dependencies, skills gaps and team motivation.
AI-written summary · about
chapters (17)
- 0:00Introduction
- 1:20We all make decisions
- 1:56Every choice has a ripple effect
- 2:41Strategies for making decisions
- 2:55Karinas background
- 4:11First story
- 6:50Second story
- 9:20Question
- 11:20Bias
- 12:24The 5 factors
- 12:47Potential
- 15:15Risks
- 16:52Involvement
- 18:51Skills
- 20:51Motivation
- 22:21Summary
- 23:10Conclusion
related talks
35mWhat Failure Taught Me About Building High-Stakes Models
31mBuilding the Business case for ML Platforms
24mFoundations for ML at Scale: Where to Start in Your Organization
43mEngineering Platforms Evolved
41mChallenges in incorporating ML models
35mBuilding Data-Intensive Teams
30mOperationalizing Machine Learning
30mBuilding Teams and Culture that Support ML Innovation