Cloudera Machine Learning Engineer study plan
Cloudera Machine Learning Engineer · Give each week a clear focus.
8 weeks · About 6 hours per week
Machine Learning Workbench and Project Setup
Create a Cloudera AI project from a Git repository and verify that the project appears in the workspace with the expected files.
Configure a Python runtime with the required libraries and confirm the environment starts successfully.
Launch a JupyterLab session and run a notebook that prints the Python version and lists installed packages.
Identify where project files, jobs, and sessions are managed in the Cloudera AI interface.
Your checkpoint
A one-page annotated screenshot guide showing project creation, runtime configuration, and a successful notebook execution.
Machine Learning Workbench and Project Setup & Data Access, Ingestion, and Preparation
Organize a project directory structure with folders for data, notebooks, scripts, and models.
Connect to a Hive table or Impala table from a session and preview the schema and first rows.
Read data from HDFS and object storage into a pandas DataFrame using the appropriate library.
Document the data access method and any credentials or connection settings in a README.
Your checkpoint
A project README that lists each data source, the access method used, and a sample query or read operation for each.
Data Access, Ingestion, and Preparation
Profile a dataset for missing values, outliers, and data types, and record the findings.
Apply at least three cleaning transformations such as imputation, type casting, and filtering.
Engineer at least two new features and justify each with a short rationale.
Write the cleaning and feature engineering steps as reusable functions in a Python module.
Your checkpoint
A data preparation notebook plus a Python module that transforms raw data into a cleaned feature table, with a summary of before-and-after row counts.
Model Development and Experimentation
Train at least three candidate models with different algorithms on the prepared dataset.
Evaluate each model with appropriate metrics and record the results in a comparison table.
Track experiment parameters, metrics, and artifacts using an experiment tracking tool available in the environment.
Select a final model and document the selection criteria.
Your checkpoint
An experiment tracking report showing at least three runs with parameters, metrics, and the chosen model artifact.
Model Development and Experimentation & Model Deployment and Serving (Cloudera AI)
Package the selected model with a preprocessing pipeline so that raw input can be scored directly.
Write a scoring script that loads the model and returns predictions for a sample input.
Create a model build in Cloudera AI and register the model artifact with a version number.
Test the registered model by invoking it with sample data and verifying the output format.
Your checkpoint
A registered model version in Cloudera AI with a test invocation log showing correct predictions for at least five sample records.
Model Deployment and Serving (Cloudera AI)
Deploy the registered model as a real-time inference endpoint and confirm it is reachable.
Send test requests to the endpoint using curl or a Python client and validate the response schema.
Configure authentication and access control for the endpoint according to a defined policy.
Document the endpoint URL, request format, and response format in a usage guide.
Your checkpoint
A deployed model endpoint with a usage guide that includes a working request example and expected response.
Operations, Monitoring, and Governance of ML Jobs
Create a scheduled job that runs the data preparation and scoring pipeline on a recurring schedule.
Configure job retries, timeouts, and alerting for failures.
Review job run history and identify at least one failure scenario and how it would be diagnosed.
Set up logging so that job output and errors are captured for later review.
Your checkpoint
A scheduled job with a run history showing at least two successful runs and one documented failure and resolution.
Operations, Monitoring, and Governance of ML Jobs & Machine Learning Workbench and Project Setup
Implement a model monitoring check that compares recent predictions against a baseline distribution.
Define a governance checklist covering data lineage, model ownership, and access review.
Review the full project for reproducibility by rebuilding the environment from the repository.
Complete a timed practice assessment covering all five domains and review incorrect answers.
Your checkpoint
A governance and monitoring checklist plus a practice assessment score report with a list of weak areas to revisit.
Milestones
Week 4: You can create a project, access data from two different sources, prepare a clean feature table, and produce an experiment tracking report comparing at least three models.
Week 6: You have a registered model version deployed as a real-time endpoint and can demonstrate a successful request-response cycle with authentication.
Week 8: You have a scheduled job with monitoring and a governance checklist, and you scored at least 70% on a timed practice assessment covering all domains.