This is an example source for a simple mlops flow implementation. It includes the following features:
- Upload data
- Data preprocessing
- Detire modeling and testing and distribution
- New data prediction
You can do a simple test through the Makefile in the root folder.
- make run-data : data implements
- make run-preprocess : data process
- make run-train : data trainning
- make run-api : data prediction, run flask app to predict through web apps.
run-data -> run-preprocess -> run-train -> run-api
- install file-api-server used "mayth/simple-upload-server"
- install ftp server used vsftpd
- install jenkins https://www.jenkins.io/doc/book/installing/
- install docker for docker-registry : https://docs.docker.com/registry/ for docker-client : https://docs.docker.com/get-docker/
- install kubernetes refer any site
- install mlflow-server in kubernetes run docker/server