Exam DP-100: Designing and Implementing a Data Science Solution on Azure
Proctored Microsoft certification exam focused on Azure machine learning solution design, implementation, deployment, and monitoring.
- Type
- Written
- Duration
- 100 min
Exam sections
Design and prepare a machine learning solution
Covers the setup of an Azure Machine Learning workspace, including compute specifications, datastores, data assets, and environment management. It also involves version control with Git and sharing assets via registries across multiple workspaces.
Preparation tips
Focus on workspace creation, compute management, and datastore configuration using the Azure ML SDK and portal.
Explore data and run experiments
Focuses on Automated ML for tabular, vision, and NLP data. Key tasks include notebook-based data wrangling, Synapse Spark integration, MLflow tracking, and automating hyperparameter tuning through specific sampling and search space configurations.
Preparation tips
Master Automated ML settings and learn to use MLflow for tracking experiment metrics and parameters in Jupyter notebooks.
Train and deploy models
Emphasizes training scripts, custom pipeline components, and model management. This includes creating and monitoring training pipelines, defining MLmodel signatures, and deploying models to online or batch endpoints for scalable inference.
Preparation tips
Understand how to define pipeline components and deploy models to managed online endpoints. Review responsible AI principles for model assessment.
Optimize language models for AI applications
Addresses Large Language Model (LLM) optimization, including prompt engineering and Retrieval-Augmented Generation (RAG). Tasks involve using Prompt Flow SDK, configuring vector stores in Azure AI Search, and fine-tuning base models with custom datasets.
Preparation tips
Experiment with the Prompt Flow SDK and understand the chunking and embedding processes required for building RAG solutions on Azure.
