I presented this talk at the Melbourne Data Engineering meetup, on a wild and wet Friday night.
Having your cake and eating it too – Effective topologies for Data and ML teams (slides)
In the talk I explore how Team Topologies provides patterns for reconciling fast flow of value with (multiple) specialisations in data and ML. I set the framework, explore examples of the 4 team shapes and their typical interaction modes, and provide some tactics and strategies for improvement. Also includes much cake.
[I later talked about this content with Matthew Skelton and David Tan on the Team Topologies Stream of Teams podcast. We covered a lot of ground, including CD4ML, the building blocks above, and of course typical shapes, interaction modes and evolution of teams within an AI/ML product delivery organisation. We had a great conversation about the role and evolution of enabling teams in particular.]
At the meetup, I also discussed the perspective of 7 wastes of data production – when pipelines become sewers, as providing a deeper analysis of common wastes in data initiatives and how they can be addressed.
The talk is based on the content in chapter 11 of Effective Machine Learing Teams; the chapter that wrote itself when we realised we needed to address between-team issues for truly effective teams.