Skip to main content

Learn Data Engineering

Data engineering, the way a data engineer actually learns it

Start with SQL, then work outward to the cloud platform, the pipelines, and the data work AI systems run on. Every concept is learned the same way: read it, write it, and practice it against a real database that runs entirely in your browser.

ReadApplyPractice

SQL

The query language every data engineering interview opens with, from your first SELECT through joins, aggregation, window functions, and the patterns companies actually ask for.

  1. Level 1: SQL Foundations (Reading Source Data)

    SELECT, WHERE, ORDER BY, types: querying raw source tables the way a DE does on day one.

    11 lessons · ~3h

  2. Level 2: Aggregation & Joins (Combining Source Data)

    Aggregates, GROUP BY/HAVING, every join flavor, subqueries, CTEs: building metrics.

    11 lessons · ~5h

  3. Level 3: Data Modeling & Schema Design

    DDL, keys, constraints, and normalization: designing schemas the database enforces.

    12 lessons · ~6h

  4. Level 4: Data Engineering with SQL

    Window functions, recursive CTEs, SCD, idempotent merge, data-quality: warehouse transforms.

    15 lessons · ~8h

  5. Level 5: Advanced & Company-Specific SQL for DE Interviews

    Rehearse the four rounds of a Data Engineering loop as graded SQL: the power screen, the modeling round, the pipeline round, and the system-design round.

    14 lessons · ~8h

Cloud and Data Platforms

The platform the SQL runs on: object storage, file formats, partitioning, warehouses, and the lakehouse, each taught by querying a simulated platform's own metadata.

  1. Level 6: Cloud & Data Engineering Foundations

    Learn the platform your SQL runs on: object storage, columnar files, partitioning, and distributed pipelines, the way a DE interview asks about them. It is beginner-friendly and can be started right after Level 2, since its graded queries lean on GROUP BY and joins.

    14 lessons · ~6h

  2. Level 7: Warehouses, Lakehouse & Dimensional Modeling

    Go inside the systems the lake feeds: how a warehouse places and prices your data, how to answer the dimensional modeling questions interviews actually ask, and how Iceberg turns files into transactional tables.

    13 lessons · ~6h

Pipelines and Reliability

Moving data on a schedule without corrupting it: orchestration, idempotency, incremental loads and backfills, streaming, and change data capture.

  1. Level 8: Batch Pipelines & Orchestration

    Run a transform every day without corrupting the table: orchestration and task states, the three idempotent load patterns graded by a double run, incremental loads and backfills, and the dbt-shaped workflow.

    13 lessons · ~6h

  2. Level 9: Streaming & Change Data Capture

    Drop the daily-batch assumption: the log and its offsets, change data capture and ordering, stream windows and watermarks computed in SQL, and how to choose batch or streaming and prove your freshness.

    12 lessons · ~5h

Compute and Operations

Distributed execution and the operational half of the job: diagnosing slow jobs, data quality monitoring, incident triage, cost, and access governance.

  1. Level 10: Distributed Compute & Data Operations

    Read a distributed engine the way an interviewer expects you to: diagnose slow jobs from Spark metrics, run production quality and observability monitors, and audit cost and governance with SQL.

    14 lessons · ~6.5h

Data Engineering in the AI Era

The pipelines AI systems run on: retrieval indexes, embedding freshness, curation and deduplication, and the semantic layer that makes generated SQL trustworthy.

  1. Level 11: Data Engineering for AI

    The data work AI systems actually run on: deduplicating and scrubbing a training corpus, measuring a curation funnel, and owning the semantic layer that decides whether generated SQL can be trusted.

    5 lessons · ~2.5h