We need a Data Platform!
summary
Argues for proving value fast with a single measurable use case (five selection rules), builds a use‑case‑agnostic domain layer to avoid duplicate logic, and implements ingestion, transformation and orchestration using DLT, dbt, Dagster on BigQuery/GCP. Delivers the first use case end‑to‑end (inventory forecasting) and gets an enthusiastic inventory manager as a user, but adoption stalls — one effective proof of value exists while the platform still lacks organization‑wide uptake.
AI-written summary · about
chapters (10)
- 0:00Building a data platform
- 2:14Why we need it
- 3:58Pitfalls in data projects
- 7:11Quick-win use cases
- 15:44Domain layers for clean models
- 25:44Open-source tools choice
- 36:28DLT, dbt, Dagster example
- 37:09Ensuring platform adoption
- 39:15Summary for success
- 41:37Answering audience questions
related talks
41mModular & Reusable: A Recipe to Scaling Data Across a Portfolio
43mScaling Data Stacks: How To Build a Great Data Model
37mBuilding a Data Platform from scratch
41mModern Data Stack: Optimising and Scaling Data in a Tech Company
28mPractical steps to establish and scale your data & ML team
56mData Unleashed: A Developer's Perspective on Navigating the Architecture Maze
43mAnalytics capabilities architectural patterns: A journey for data-driven insights
39mData Mesh: Data Analytics for Engineers