The Work Gym

Courses / Data and analytics tools

dbt

Build, test, and document analytics projects in a working dbt-style IDE.

Start lesson 1 Practice interview Take the assessmentPractice the first unit of every course. No account needed.

52lessons

342checked exercises

13projects, cases and exams

Prepares you for Analytics engineering

Projects, case studies and mock exams

Project · BeginnerA shop staging layer

Harbor Cart's merchandising analyst receives duplicate order updates and inconsistent contact codes.

6 steps · 18 min Case study · BeginnerA late loader

Harbor Cart's morning report looks old.

6 steps · 15 min Project · IntermediateAn order reporting mart

Harbor Cart finance needs a report that can survive a source correction without changing its definitions.

6 steps · 18 min Case study · IntermediateConflicting subscription totals

Pulse Path's operations and finance teams report different active-account totals.

6 steps · 15 min Project · AdvancedReplay-safe order updates

Deliver a current-order model, a two-day replay contract, checks, and currency-aware totals.

6 steps · 28 min Case study · AdvancedA plan history dispute

Build a snapshot and checked evidence for an observed subscription plan change.

6 steps · 25 min Project · AdvancedA documented subscription mart

Pulse Path revenue operations needs a small dataset another analyst can adopt without a meeting.

6 steps · 18 min Project · ExpertA safe model release

The Harbor Cart reporting team wants a smaller release that standardizes customer region codes without rebuilding unrelated work.

6 steps · 18 min Case study · ExpertA failed reporting release

A Harbor Cart report changed after a release that was meant to clean text codes.

6 steps · 15 min Mock exam · ExpertCertification-style scenario mock

This independent practice exam contains 65 original scenario questions with a 120-minute time limit.

65 questions · 120 min

Practice analytics engineering in a browser project that compiles Jinja, builds DuckDB models, and checks your results. Start with SQL models and source declarations, then develop reusable staging layers, tests, and reporting marts. Work through original fictional shop and subscription data with duplicates, missing fields, and late corrections. The complete curriculum progresses through incremental processing, history, packages, contracts, and release workflows, with connected projects, incident cases, interviews, and timed certification practice. No local warehouse account is needed.

Projects that use dbt

Work one company brief through several tools. Your dbt stage reads the tables from the stage before.

Project · WayfriendFrom invites to meetupsDatabricks → dbt → SQL → Dagster · 3 hr 30 min Journey · WayfriendYour first 90 days as a data engineerDatabricks → dbt → SQL → Dagster · 8 hr

See all projects

Beginner

Project foundations

Build named models, configure relations, and connect dependencies.

4 lessons

1ELT and your first modeldbt turns a SQL SELECT into a named dataset in your warehouse.4 exercises · 12 minFree

2Project files and configurationA dbt project keeps SQL models, YAML properties, seeds, tests, snapshots, and macros in separate folders.4 exercises · 12 minFree

3References and the dependency graphA ref call links a model to another model in the project.4 exercises · 12 minFree

Practice setProject foundations practiceReview model projections, source records, and NULL-safe SQL.Free

Sources and relations

Declare warehouse inputs and choose how transformed relations are stored.

4 lessons

5Loaders and source freshnessA loader such as Fivetran or Airbyte copies records into a warehouse.4 exercises · 12 minWith a plan

6Source declarations and raw recordsA source declaration gives a raw relation a stable logical name.4 exercises · 12 minWith a plan

7Views, tables and ephemeral modelsA view stores a query definition, while a table stores the query's result.4 exercises · 12 minWith a plan

Practice setSources and relations practiceReview source boundaries, physical relations, and deliberate cleanup.With a plan

Build and test

Run selected branches, test guarantees, and diagnose an upstream delay.

5 lessons

9Commands and node selectiondbt commands have different jobs.4 exercises · 12 minWith a plan

10Generic data testsA data test is a query that returns records violating a rule.4 exercises · 12 minWith a plan

ProjectA shop staging layerHarbor Cart's merchandising analyst receives duplicate order updates and inconsistent contact codes.With a plan

Case studyA late loaderHarbor Cart's morning report looks old.With a plan

Practice setBuild and test practiceReview source inspection, model grain, and operational command choices.With a plan

Intermediate

Analytics models

Build reusable layers and a tested reporting mart.

5 lessons

14CSV seeds and reference dataA seed is a small CSV file versioned with the project.4 exercises · 12 minWith a plan

15Staging, intermediate and martsLayers make a project easier to change without repeating cleanup rules.4 exercises · 12 minWith a plan

16Singular tests and failing rowsA singular test is a SQL file under tests that returns the rows violating a business rule.4 exercises · 12 minWith a plan

ProjectAn order reporting martHarbor Cart finance needs a report that can survive a source correction without changing its definitions.With a plan

Practice setAnalytics models practiceThese projects use maintained lookups, revisioned exports, and operational line items.With a plan

Jinja and reuse

Generate readable SQL with reusable macros, variables, and isolated targets.

5 lessons

19Jinja expressions and control flowJinja runs before the warehouse executes SQL.4 exercises · 12 minWith a plan

20Macros and argumentsA macro gives a repeated Jinja pattern a name.4 exercises · 12 minWith a plan

21Variables and deployment targetsProject variables let a model use a named setting without hardcoding the value in every file.4 exercises · 12 minWith a plan

Case studyConflicting subscription totalsPulse Path's operations and finance teams report different active-account totals.With a plan

Practice setJinja and reuse practiceA template can produce valid SQL and still encode the wrong rule.With a plan

Data quality and documentation

Express scoped data guarantees and document downstream ownership.

5 lessons

24Test severity and row filtersA test can carry configuration as well as a predicate.4 exercises · 12 minWith a plan

25Custom generic testsA custom generic test packages a failing-row query so you can apply it to more than one model or column.4 exercises · 12 minWith a plan

26Descriptions, docs and exposuresDescriptions explain what a dataset means at its declared grain.4 exercises · 12 minWith a plan

Interview setEasyAn analytics engineering interview tests whether you can explain a choice and its consequences.With a plan

Practice setQuality and documentation practiceThe quality tasks start with business rules that need different tests.With a plan

Advanced

Incremental processing

Process new arrivals and corrections with explicit keys and replay behavior.

5 lessons

29Incremental append modelsBuild an append table that handles first runs, arrival clocks, and repeated batches.4 exercises · 18 minWith a plan

30Incremental merge and unique keysReplace corrected keys, read late updates, and keep one deterministic incoming row per key.4 exercises · 18 minWith a plan

31Delete and insert, then full refreshReplace incoming keys and rebuild historical rows when the normal batch cannot reach them.4 exercises · 18 minWith a plan

ProjectReplay-safe order updatesDeliver a current-order model, a two-day replay contract, checks, and currency-aware totals.With a plan

Practice setIncremental processing practiceRepair empty watermarks, partial batches, late corrections, source deletions, and changed model definitions.With a plan

Historical records

Record observed changes and distinguish history from current state.

4 lessons

34Timestamp snapshotsRecord observed source changes as closed and open validity intervals.4 exercises · 16 minWith a plan

35Check snapshots and change columnsCompare business fields, including nullable values, without recording unrelated reloads.4 exercises · 16 minWith a plan

Case studyA plan history disputeBuild a snapshot and checked evidence for an observed subscription plan change.With a plan

Practice setHistorical records practiceApply snapshot keys, comparison scopes, NULL changes, replay checks, and interval boundaries.With a plan

Packages and guarantees

Reuse package macros and declare execution and schema guarantees.

4 lessons

38Packages and dbt_utils patternsA package supplies reusable project code.5 exercises · 15 minWith a plan

39Hooks and model contractsA model contract checks the declared output shape, including column names and types, when the adapter supports it.4 exercises · 12 minWith a plan

ProjectA documented subscription martPulse Path revenue operations needs a small dataset another analyst can adopt without a meeting.With a plan

Practice setPackages and guarantees practiceThese tasks use package macros to build output, then add guarantees around that output.With a plan

Expert

Diagnosis and maintenance

Repair compilation defects and evaluate refactors without changing semantics.

4 lessons

42Compilation errors and safe refactoringCompilation happens before dbt can send model SQL to the warehouse.4 exercises · 12 minWith a plan

43Materialization and query performancePerformance work starts with a correct result and a measured bottleneck.4 exercises · 12 minWith a plan

Interview setMediumAn analytics engineering interview tests whether you can explain a choice and its consequences.With a plan

Practice setDiagnosis and maintenance practiceThe failure stage tells you where to look.With a plan

Delivery workflows

Validate changed resources with state-aware workflows and a stable consumer contract.

4 lessons

46State selection and continuous integrationA state comparison lets a job select resources that differ from a saved project artifact.4 exercises · 12 minWith a plan

ProjectA safe model releaseThe Harbor Cart reporting team wants a smaller release that standardizes customer region codes without rebuilding unrelated work.With a plan

Interview setHardAn analytics engineering interview tests whether you can explain a choice and its consequences.With a plan

Practice setDelivery workflows practiceThe shipment project gives you a concrete dependency graph and a saved list of changed files.With a plan

Assessment

Resolve a release incident and practice original certification-style decisions.

3 lessons

Case studyA failed reporting releaseA Harbor Cart report changed after a release that was meant to clean text codes.With a plan

Mock examCertification-style scenario mockThis independent practice exam contains 65 original scenario questions with a 120-minute time limit.With a plan

Practice setExpert transfer practiceThe final tasks combine transformation logic with project guarantees.With a plan