Steering a data science project
summary
Checklist named Portillo guides project direction with concrete choices: an autoencoder embeddings pipeline creates image vectors and sequences, a user feedback loop captures corrections and brand uploads for retraining, and teams split models — one model per product category — to bootstrap accuracy. Tests use pytest and Git/GitHub with CI, and behavior-driven specs document scenarios. Risk assessment follows ISO 42001 guidance. The project ships a working pseudo-classification sequencer that improves incrementally.
AI-written summary · about
chapters (11)
- 0:00Introduction
- 2:09Checklist essentials
- 3:55Project vision and problem
- 6:25Team communication in AI
- 7:33Testing AI systems
- 10:09Usability in AI solutions
- 13:50Data and model lifecycle
- 16:19Progress with checklist
- 17:45Risk identification
- 19:33Q&A: Lifecycle deep dive
- 22:00Q&A: Tool selection
related talks
40mBuilding Successful Data Science Projects
16mKey Points and Lessons Learned in AI Projects in Mechanical Engineering
31mProductionalizing a Data Science Team
40mMistakes were made: Data science ten years in
25mFrom Zero to AI Hero
54mThe secret sauce of data science management
42mDevelop Data Science for Business Value: a Phoenix Story
21mApplied AI & Data Analytics