Data Science – The New Troublemaker in the Delivery Pipeline
summary
Claims that data science must automate operational decisions and be measured against real euro value; it needs raw, near‑real‑time inputs, common IDs, fine‑grained access control and resilience. Examples show a small retailer with 100 stores and 3 years and 50 features exploding into 500 million rows for training and producing around 7 million predictions per day, which requires heavy compute, changed hardware like digital price tags, cross‑functional teams, continuous validation and a clear ROI before scaling.
AI-written summary · about
chapters (15)
- 0:00Introduction
- 1:12What is data science
- 4:03Why ops decisions matter
- 8:25Optimal decisions in retail
- 10:39Why integration is hard
- 13:58How DS disrupts processes
- 20:27Data science’s insatiable appetite for resources
- 25:12Security and compliance hurdles
- 27:02DevOps for data science
- 31:04Simplify DS models
- 32:52Data access vs security
- 34:21Modernize IT for DS
- 36:04Roadmap for DS integration
- 37:33Key takeaways
- 38:38Q&A
related talks
19mData science demystified
14mData-Driven Transformation: Scaling Businesses with AI, Ethics, and Strategic Vision
36mBig Data, Disruption And The 800 Pound Gorilla In The Corner
29mBecoming a data-driven enterprise
26mOperating Data Science: Turning Rocket Science into Real Money
43mDomain-driven Data Science
25mScience Will Eat Its Children
25mBuilding A Successful Data Science Team