The Professional Data Engineer exam tests whether you can design, build, store, serve and operate data systems on Google Cloud, choosing the most managed service that meets the stated constraint. This outline lists what each domain expects, in plain terms. Tick topics off as you cover them. Your ticks are saved in this browser.
This outline follows the official PDE exam guide (PDF). Check it for the exact wording.
Designing data processing systems 22% Designing secure, reliable, portable data systems and planning migrations. About 22% of the exam.
Security and compliance IAM roles for data services, service accounts, and organization policies as guardrails Encryption and key management: Google-managed keys, CMEK, Cloud HSM, Cloud EKM Handling PII with Sensitive Data Protection (Cloud DLP): inspect, mask, tokenize, date-shift Data sovereignty: choosing regions and multi-regions, resource location constraints, Assured Workloads Meeting legal and regulatory needs: retention, audit logs, erasure, VPC Service Controls Organising projects, datasets and tables so governance is enforceable Separating development and production environments and promoting between them Reliability and fidelity Cleaning and preparing data with Dataform, Dataflow, Cloud Data Fusion and LLM-generated SQL Monitoring and orchestrating pipelines Disaster recovery and fault tolerance: RPO, RTO, replication, time travel, snapshots Choosing between ACID guarantees and availability: Spanner, Cloud SQL, Bigtable, BigQuery Validating data with assertions, quality scans and dead-letter sinks Flexibility and portability Data migrations
Ingesting and processing data 25% Planning, building and operating batch and streaming pipelines. The largest domain, about 25%.
Planning pipelines Building pipelines Data cleansing: nulls, types, deduplication, quarantining bad records Choosing among Dataflow, Beam, Dataproc, Data Fusion, BigQuery, Pub/Sub, Spark, Hadoop and Kafka Batch transformations Streaming transformations: windows, triggers, watermarks, late data, exactly-once Processing logic: joins, side inputs, state, aggregation AI data enrichment with BigQuery ML remote models, embeddings, RunInference and AI APIs Data acquisition and import: load jobs, Storage Write API, subscriptions, DTS, Datastream Integrating new data sources with managed connectors, CDC and custom ingestion Deploying and operationalising
Storing data 20% Choosing storage, modelling warehouses, running lakes and building governed platforms. About 20%.
Selecting storage systems Analysing access patterns before choosing a store BigQuery, BigLake, AlloyDB, Bigtable, Spanner, Cloud SQL, Cloud Storage, Firestore and Memorystore compared Planning storage cost and performance: classes, Autoclass, SSD vs HDD, long-term storage Lifecycle management: retention, expiration, versioning, archiving Data warehouse design Data lakes Data platforms
Preparing and using data for analysis 15% Making data fast, safe and shareable for BI, AI and partners. About 15%.
Preparing data for visualisation Preparing data for AI and ML Sharing data
Maintaining and automating workloads 18% Cost, capacity, automation, monitoring and recovery. About 18%.
Optimising resources Automation and repeatability Workload organisation Monitoring and troubleshooting Failure awareness and mitigation