DEV Community

Cover image for MLOPs Lifecycles

Posted on

MLOPs Lifecycles

MLOPs life cycle ๐Ÿ”ƒ โคต

  • Define business need

  • Getting datasets ready. This phase includes data cleaning, labelling, pixel optimisation, open source datasets and enterprise data-lakes

  • Model development (code part)

  • Training and optimisations of model

  • Deployment(uat/prod)

  • User interface and API development to let user interact with the model

  • Monitoring both model and system resources

  • Insights and analytics

  • Continuous model training: deployed one time model can work for few time-frames and hence it needs retraining on new data

  • world has changed. We use CI/CD tools like Jenkins and Docker/cubectl for code automation

  • Security and protection of ML model against known vulnerability. This is the point most people ignore


Top comments (0)