MLOps is messy at the moment, but the value it can deliver to the society is big

MLOps is messy, but the value it can deliver for society is big, writes Michail Eric in his blog. Shortly he observes the following trends:

1. Closing the loop in ML systems
2. Declarative systems for ML
3. Real-time -ML
4. Mergining business insight tooling with DS/ML workflows
5. Better data management

and the following META-trends

1. Talent shortage
2. Increasing consolidation around E2E platforms
3. Cultural adoption of ML thinking (similar to DevOps): holistic systems thinking (break silos), augmenting feedback loops in products, and experimentation and learning.