MLflow

MLflow provides experiment tracking, model versioning, and artifact management for ML workloads.

Architecture

+------------------+     +------------------+
| MLflow Tracking  |     | PostgreSQL       |
| Server           |---->| (Backend Store)  |
| Port: 5000       |     +------------------+
+------------------+
       |
       v
+------------------+
| MinIO / S3       |
| (Artifact Store) |
+------------------+

Configuration

# From ml-service values
config:
  mlflow:
    trackingUrl: "http://mlflow:5000"

MLflow stores experiment metadata in PostgreSQL and model artifacts in S3/MinIO.

Key Features

Feature	Description
Experiment Tracking	Log parameters, metrics, and artifacts per run
Model Registry	Version, stage, and promote models
Artifact Store	S3-compatible storage for model files
REST API	Programmatic access from ML Service

Overview Feast