Evolution Of ML Use Cases

Created: 2023-02-13 11:06
#note

1. Exploration -> small team which studies the technology ecosystem, scouts use cases, and collect sample data. They also try modeling and they check if the project is doable;
2. Experimentation -> bigger team which selects a use case, collects and processes training data, conducts training experiments, reviews results, and finally decides on go/no go;
3. Engineering -> full-fledged team, which decides requirements, designs ML flow, builds complementary software, trains, tests, integrates, and deploys.

MLOps can not be used to soon because it is time consuming, but not too late or difficulty of the project could lead to problems.
So, we should start use it incrementally in phase 2, and then in phase 3.

Tags

#mlops #ml