Final Exam: MLOps

WHAT YOU WILL LEARN

• 

  Identify the features of mlops
  

  outline the ml workflow
  

  outline level 1 and 2 of ml pipelines
  

  provide an overview of mlflow
  

  outline the use of the machine learning (ml) workflow
  

  recognize the features offered by mlflow
  

  outline how mlflow signatures work
  

  load, clean, and visualize data for machine learning
  

  view statistics about data with pandas profiling and use it to view correlations
  

  create an mlflow experiment and explore it using the mlflow user interface (ui)
  

  create a run using a with block and view run info
  

  compare mlflow models using the ui
  

  visualize and clean data
  

  run a classification model and view the metrics
  

  register an mlflow model
  

  serve models to a rest endpoint and access that endpoint
  

  deploy a model to azure and view this model
  

  outline how mlflow works with databricks
  

  create a search space and define a search algorithm
  

  run a hyperparameter tuning model and view the results
  

  run a hyperparameter tuning model and view the results
  

  clean data for a time-series model
  

  create a machine learning model and set up model evaluation
  

  run a model and evaluate that model
  

  train an image classification model and run it
  

  outline the use of pytorch and view images for machine learning
  

  run a sentiment analysis model and view logged artifacts
  

  outline how to work with mlflow projects
  

  create an experiment for a project, run it, and view results
  

  outline the use of mlflow recipes
• 

  modify the train.py file in a recipe and modify the custom_metrics.py file
  

  validate models based on a metrics threshold
  

  train using data from databricsk file system (dbfs) and delta lakes
  

  outline key concepts of dvc
  

  describe the features of dvc
  

  create a git local repository
  

  create and serialize an ml model
  

  run and push a different model version
  

  set up an ml pipeline stage
  

  execute a dvc pipeline
  

  set up dvc in a jupyter notebook
  

  visualize data for ml
  

  track ml experiments using dvclive
  

  run and track a k-nearest neighbor (knn) regression model
  

  compare logistic regression models
  

  register a classification model with the iterative studio registry
  

  outline key concepts of mlem
  

  install and set up docker
  

  get predictions from endpoints hosted on docker
  

  connect to an s3 bucket from dvc
  

  set up a convolutional neural network (cnn) for image classification
  

  improve the image classification model
  

  track tensorflow models in dvc
  

  set up a data version control (dvc) project for a machine learning (ml) pipeline
  

  run a pipeline stage
  

  add an evaluation stage to an ml pipeline
  

  run dvc experiment pipelines
  

  outline key concepts of cml
  

  set up a cml workflow for continuous integration and continuous delivery (ci/cd)
  

  view performance reports with every git push