Learner track

Over the next couple of hours you take a working NYC-taxi app and rebuild its data plane on ClickHouse Cloud: real historical analytics, a live change-data-capture stream, conversational BI, an AI-built SRE dashboard, and an in-app AI chat traced end to end. Work the modules in order - the numbered list below doubles as your progress tracker.

Your progress

0. Setup - accounts, keys, tools, agent skills, and the app running locally.
1. ClickHouse Cloud - build the schema and seed 3.2M+ real taxi rows from object storage.
2. Base app - tour the running app, now that ClickHouse is seeded, so you know what you are extending.
3. Real-time CDC - stream your own Postgres into ClickHouse with a ClickPipe.
4. ClickHouse Agents - ask your data questions in plain language.
5. ClickStack - see the app's own traces and logs in HyperDX.
6. AI SRE - have your agent build an SRE dashboard and alert over that telemetry.
7. Break and fix - continue from AI SRE or optional 06b, inject a fault, diagnose it, and fix it.
8. Chat and Langfuse - wire an in-app AI chat and trace every turn.
9. Wrap-up - recap, take it home, extend it to your own data.

What you'll build

By the end you have, all running on your own trial service:

• a live ops dashboard over 3.2M+ real NYC taxi rows seeded straight from object storage,
• a change-data-capture pipeline streaming your own Postgres into ClickHouse continuously,
• conversational BI over that data with ClickHouse Agents,
• an AI-built SRE dashboard and alert over the app's own telemetry,
• and an in-app AI chat traced end to end in Langfuse.

Time

About 2h30 hands-on - and your dashboard is live on real data within the first 40 minutes.

Cost

$0 - ClickHouse Cloud trial credits cover everything. The only paid key is about $5 of OpenAI credit for the chat module (module 08), and even that stretches across many runs.

Everything is copy-paste

Every command and query on these pages is ready to copy and run, and your coding agent does the heavy lifting - you are directing the build, not typing it out.

How each lesson is structured

Every module page follows the same shape, so you always know where you are:

1. Starting point - the checkpoint branch to check out and what you need in place.
2. Why - why this step matters for the app you are building.
3. Goal - the one concrete outcome for the module.
4. Steps - numbered, each verb-first, with the exact files you touch and an expected-output checkpoint.
5. How to verify you are done - a check you can run yourself.
6. Wrap-up - what just happened.
7. End state - the handoff into the next module.

Do not worry about keeping pace with everyone else: every module names a checkpoint branch you can check out to start fresh, so you are never stranded.

Working on your own

Running this without an instructor? Read Running this workshop self-paced first - it maps what an instructor would otherwise provide onto the self-serve replacements built into these materials. If anything breaks, Troubleshooting lists every setup failure seen in testing with its fix.

Start with 00 Setup.