Apache Kafka · from first event to FAANG-level design

Stop saying "just put it on Kafka".
Know the key, the partitions, and what breaks.

Kafka taught as distributed logs, partitions, replication, ordering, consumer coordination, delivery semantics, storage and failure handling, not as a list of APIs. Every topic works through a real interview problem and the ways it breaks in production: keys and partition counts, acks and min.insync.replicas, idempotent producers, rebalancing, exactly-once and its limits, retry topics and DLTs, schemas, Connect and CDC, the outbox and inbox, Kafka Streams, KRaft, security, capacity planning, multi-region DR, sagas, event sourcing and CQRS, ending with staff-level designs, a capstone and a full question bank.

Start from zero The full roadmap

Phases: 18
Topics: 87
Hours: 11
Diagrams: 42
Exercises: 302

By the end you can

What this course makes you able to do

Choose topics, keys and partition counts from ordering and throughput requirements, and handle hot partitions
Configure producers and topics so acknowledged events survive broker and AZ failures without duplicates
Build consumers with correct commits, rebalancing, retries, retry topics and DLTs, and idempotent processing
Explain exactly what exactly-once covers, and use transactions, the outbox and the inbox where they belong
Run Kafka in production: KRaft, security, quotas, capacity planning, monitoring, debugging and multi-region DR
Design event-driven systems (sagas, event sourcing, CQRS, CDC, analytics pipelines) with the full reasoning chain

Beginner → Advanced

The curriculum, by level

Beginner

Intermediate

Advanced

Where should I start?

Pick a path for your goal

Backend developer (0–2 years)

You've published or consumed a few events and want to do it properly.

Then Phase 4 to understand what really makes an event durable.

Interview-ready (2–4 years)

You'll design event pipelines and explain delivery guarantees in interviews.

Finish with the Phase 17 question bank.

Senior / Staff

You own Kafka platforms or lead event-driven architecture decisions.

Helpful before this course

System Design · Messaging

Where queues, logs and event-driven services fit in a whole architecture.

Docker course

Every lab in this course runs Kafka in KRaft mode with Docker Compose.

Where this leads next

Stateful Systems · Kafka field guide

Operating Kafka next to PostgreSQL and Redis: labs, knobs, and failure drills.

Observability course

Metrics, logs, and traces for the lag and latency incidents in Phase 14.

SRE course

SLOs, capacity, and incident response for event-driven platforms.

Command Palette

Stop saying "just put it on Kafka".Know the key, the partitions, and what breaks.