Make MLOps great Again! A painless, Cloud-Native approach with MLflow and Docker

In the realm of MLOps, the challenge of streamlining model deployment is crucial. Our presentation addresses this challenge with a focus on simplifying the developer experience. Deploying machine learning models is increasingly complex, and heavy-duty tools with a steep learning curve, such as KubeFlow or Seldon Core, affect the developer workflow and the efficiency of the CI/CD pipeline. We present an approach that simplifies the MLOps workflow by emphasizing containerization and cloud-native best practices using MLflow, Docker, and HELM. Our approach significantly reduces deployment times. We present a model packaging format and distribution method by publishing self-contained Docker images into OCI registries like DockerHub, AWS ECR, Google GCR, or the Azure Container Registry. Zoltan C. Toth: https://www.linkedin.com/in/zoltanctoth/ Lajos Papp: https://www.linkedin.com/in/lalyos/ #ml #machinelearning #mlops #mlflow #docker #kubernetes #ai #cicd #oci #helm