Unevenly Distributed #6: The Primitives of a Modern Data Platform ft. Hazel Weakly (Fellow of the Nivenly Foundation)
Mar 6, 2026
Hazel Weakley, a Nivenly Foundation fellow and Haskell Foundation board member who helps open source communities scale, reframes the lakehouse as a user-centered utility. She tells a story of a failed project from missing user research. They discuss schema-attached files, dynamic assembly of heterogeneous data, balancing developer ergonomics with compliance, and how AI shifts risk and authorization needs.
AI Snips
Chapters
Transcript
Episode notes
Costly Internal Project Failed From Politics Not Tech
- Hazel recounts a $10–30M internal project where multiple teams duplicated efforts and it was canned.
- The real problem was organizational politics and migration cost analysis, not the technical solution being built.
Ship A Narrow Open Source Component To Find Real Users
- Open source a focused corner of a larger problem to attract real users and iterate quickly.
- Robert described VMware releasing a widget set on GitHub, which drove adoption and direct developer feedback across products.
Change Mindsets Not Just Rules To Unlock Value
- The biggest leverage in changing systems is shifting people's conceptual models, not small rule changes.
- Hazel looks for tensions (e.g., regulation vs. modern data platforms) where reframing enables capabilities impossible under old models.