Databricks ML Professional - Study Plan¶

6-Week Study Schedule¶

Week 1: Feature Engineering (Domain 1)¶

Day 1-2: Feature Store Architecture¶

• Understand offline vs online Feature Store
• Learn Feature Engineering Client APIs
• Practice create_table, write_table, create_training_set workflow
• Review Notes: notes/01-feature-engineering.md
• Read: Feature Store

Day 3-4: Point-in-Time Lookups¶

• Understand why point-in-time prevents target leakage
• Practice timestamp_lookup_key configuration
• Design time-series feature pipelines
• Read: Point-in-Time

Day 5-7: Advanced Feature Engineering¶

• Practice time-series features: rolling windows, lag features
• Understand feature selection methods (SHAP, mutual information, L1)
• Learn feature freshness monitoring
• Review feature versioning with Delta Lake

Week 2: MLOps Pipeline (Domain 2 - Part 1)¶

Day 8-9: Production ML Pipeline Design¶

• Design multi-step training pipelines with Databricks Workflows
• Understand idempotent pipeline design
• Learn Databricks Asset Bundles for deployment
• Review Notes: notes/02-mlops-pipeline.md
• Read: Asset Bundles

Day 10-11: CI/CD for ML¶

• Understand dev/staging/production promotion patterns
• Learn automated testing for ML pipelines
• Practice model validation gates
• Read: Workflows

Day 12-14: Model Lifecycle Management¶

• Master model aliases (champion, challenger)
• Understand webhooks for model event automation
• Practice model lineage and reproducibility
• Learn model packaging: signatures, environments, requirements
• Read: Model Registry

Week 3: MLOps and Advanced ML (Domains 2-3)¶

Day 15-16: A/B Testing and Custom Models¶

• Understand traffic splitting for model comparison
• Learn champion/challenger deployment patterns
• Implement custom pyfunc models for ensembles
• Practice load_context and predict methods
• Read: Custom PyFunc

Day 17-18: Advanced Hyperparameter Tuning¶

• Practice conditional search spaces with Hyperopt
• Learn early stopping and warm-starting
• Understand parallelism tradeoffs in SparkTrials
• Review Notes: notes/03-advanced-ml.md
• Read: Hyperopt

Day 19-21: Distributed Training¶

• Understand data parallelism vs model parallelism
• Learn TorchDistributor for PyTorch distribution
• Study DeepSpeed ZeRO stages for memory efficiency
• Practice mixed precision training concepts
• Read: Distributed Training

Week 4: Deployment and Monitoring (Domains 4-5)¶

Day 22-23: Model Serving Endpoints¶

• Configure serverless and provisioned throughput endpoints
• Practice REST API queries to serving endpoints
• Understand auto-scaling and scale-to-zero
• Review Notes: notes/04-deployment-serving.md
• Read: Model Serving

Day 24-25: Deployment Strategies¶

• Understand canary, blue/green, and shadow deployments
• Learn inference table logging and configuration
• Practice batch inference with mlflow.pyfunc.spark_udf
• Read: Inference Tables

Day 26-28: Monitoring and Drift Detection¶

• Understand data drift vs concept drift vs prediction drift
• Learn statistical tests: KS test, PSI, chi-squared
• Practice Lakehouse Monitoring setup
• Design retraining triggers and automated pipelines
• Review Notes: notes/05-monitoring.md
• Read: Lakehouse Monitoring

Week 5: Integration and Scenario Practice¶

Day 29-31: End-to-End ML System Design¶

• Design complete ML systems: features to monitoring
• Combine Feature Store, MLflow, serving, and monitoring
• Practice cross-domain scenario questions
• Review all notes for integration points

Day 32-35: Practice Scenarios¶

• Work through scenarios.md exam-style questions
• Take available practice exams under timed conditions
• Review incorrect answers and understand reasoning
• Identify and focus on weak areas

Week 6: Final Preparation¶

Day 36-38: Comprehensive Review¶

• Re-read all notes and fact sheet
• Focus on MLOps (30%) and Advanced ML (25%) - the largest domains
• Create summary of key API patterns and code snippets
• Review deployment and monitoring concepts

Day 39-42: Final Review and Exam¶

• Light review of key differentiators
• Review common pitfalls and exam tactics
• Take the exam with confidence

Study Tips¶

Time Allocation by Domain Weight¶

Recommended Daily Schedule¶

• 45 min reading documentation and notes
• 45 min hands-on practice with advanced patterns
• 15 min reviewing concepts and code patterns
• Total: ~1.75 hours/day