MLFlow Set-up For Model Results - Develop goals, what use cases do we need?

[RobotPsychologist]

In this ticket we need to better understand what the primary use cases of mlflow will be for our project.

• Read/review MLflow Overview: https://mlflow.org/docs/latest/introduction/index.html
• Read/review Getting Started with MLflow: https://mlflow.org/docs/latest/getting-started/index.html
• Read/review Traditional ML: https://mlflow.org/docs/latest/traditional-ml/index.html

What articles should we read next?

• MLflow Tracking/Projects/Models/Model Registry/Recipes/CLI/Search Runs/Search Experiments

This ticket is complete when we have read and reviewed the above articles and made a plan for which MLFlow tools we will use in our project. We should create single issue tickets for each piece that a developer can work on.

Deliverable

• a list of tasks that can correspond to a single issue ticket.

[RobotPsychologist]

@gavinkatz001 @Tony911029 @paulpark6 @aryavkin @sneha3799 @liya-zhu

We're the "wrap-up" team. The "light on exams" team. Let's make a plan around how we are going to approach this for our full modeling runs. Please read over these documents and have some thoughts. Post your comments/tech discussion here.

[gavinkatz001]

commenting to get assigned.

[Tony911029]

Maybe we need some child issues for this one?

[RobotPsychologist]

Maybe we need some child issues for this one?

This isn't a ticket for actually doing the work, this is the exploratory work. So the work for this ticket is to make a list of what the new tickets should be so that they can be assigned to individual devs.

[gavinkatz001]

Ticket 1: Set Up MLflow Tracking
Ticket 2: Standardize ML Workflows with MLflow Projects
Ticket 3: Log and Deploy Models Using MLflow Models
Ticket 4: Implement Model Versioning with MLflow Model Registry
Ticket 5: Automate Workflows with MLflow CLI

I have created plans/descriptions for each of these tickets and can start creating them once I get approval.