Develop Data Science for Business Value: a Phoenix Story
summary
Claim: only 13% in production of data science projects actually ship. Project built is Phoenix color extractor — a computer vision pipeline that starts with a 13-class deep model then iterates to simpler center-crop clustering and superpixel grouping. Measured outcomes include a latency drop from about 29 seconds to interactive response, an exposed API and monitoring endpoint, documented dataset mapping and A/B testing; the system proves usable, yields clearer color metadata, and drives incremental business value through iterative demos.
AI-written summary · about
chapters (14)
- 0:00Data science for business value
- 1:28Why projects fail to deliver
- 4:42My background
- 7:34Phoenix color ID project
- 10:07Stakeholder alignment
- 12:57Iterative development
- 17:12Minimal viable product
- 21:48Tangible demos
- 29:55Problem and solution
- 34:06Iterating and refining
- 34:28Production deployment
- 38:19Recap and benefits
- 39:44Future reading
- 40:36Q&A
related talks
54mFrom the Life of a Data Scientist
46mPanel Discussion: Business Side of Data Science
40mBuilding Successful Data Science Projects
37mBridging the gap: from Data Science to service
28mData Science Project Patterns that Work
40mMistakes were made: Data science ten years in
43mDomain-driven Data Science
35mAstrophysics to data Science: how the Milky Way is like my company.