MLOps.community Evolving Workflow Orchestration // Alex Milowski // #291
30 snips
Feb 14, 2025 Alex Milowski, a serial entrepreneur and computer scientist with a PhD in Informatics, dives into the evolving landscape of workflow orchestration. He discusses the shift from traditional workflow languages to code, emphasizing its implications for complexity in orchestration. Milowski also highlights the importance of visual tools for process modeling and the unique challenges that regulated industries face in machine learning. He contrasts modern technologies with traditional automation, stressing human oversight in AI workflows and the delicate balance of documentation in agile systems.
AI Snips
Chapters
Transcript
Episode notes
Workflow Origins
- Workflows originated in the 1920s from process engineering and manufacturing.
- Modern workflow systems evolved from rule-based expert systems in the 1990s.
BPMN in the Lab
- At MicroByre, Alex used BPMN to map scientific workflows involving humans, robots, and ML.
- This improved lab efficiency and facilitated communication across teams.
Workflow Layering
- Don't think of workflows as a single entity, but rather as layered processes.
- Consider technical and business aspects, optimizing for the larger organizational context.