Home
Concurrent for MLflow is an Apache Licensed open source project for running MLflow projects, in parallel, in Kubernetes. It enhances MLflow with two important capabilities:
- Effortlessly run MLflow Projects in Kubernetes, without the hassle of dealing with docker
- Design a DAG of MLflow Projects and then execute the DAG in Kubernetes
Concurrent for MLflow is ideal for complex pre-processing of AI data and for batch/micro-batch inferencing. It is not suitable for distributed training or real time inferencing.
Simple MLflow Project Use Case¶
- Wrap your AI code in MLflow projects and store in git
- Run MLflow Project in Kubernetes
Sophisticated Parallelized DAG Use Case¶
- Wrap your AI code in MLflow projects and store in git
- Define a DAG of MLflow projects in Concurrent using the Concurrent Web UI
- Run the DAG in Kubernetes