Practice behavior across service boundaries
Backend changes often look local while affecting another part of the system. A retry can duplicate a write, an acknowledgement can remove work owned by a different worker, and a faster data query can change ordering. These exercises ask you to trace the caller, the service, and its state before changing the implementation.
Choose a problem around a question
Start with the HTTP retry client if you want to reason about repeated requests and time budgets. Choose the lease scheduler to practice ownership and concurrent workers. Payment webhooks extend the repetition question to incoming events; document access and pagination add tenant boundaries and stable reads.
Use a familiar language for your first attempt so you can focus on behavior. Each preview lists the languages and estimated time for that exercise. Estimates are planning guidance, not a promise about how quickly you should solve it.
- Reproduce the reported symptom and run the baseline tests.
- Trace one request through the relevant repository modules.
- Review any AI proposal against the service contract and affected callers.
- Run focused regressions and the existing suite after your last change.