See what the reviewer receives.

Source register

All excerpts below are illustrative, including filenames and check results. They explain the review method and contain no real candidate data.

1. [01]

   Regression test

   Illustrative

   Reproduce the duplicate write.

   tests/test_event_processor.py

   deliver(event_id="evt-014")
   deliver(event_id="evt-014")
   assert count_rows(event_id="evt-014") == 1
   
   # Before the fix: expected 1 row; found 2.

   What it supports

   The test expresses the failure as an observable result: delivering the same event twice must create only one record.

   What remains open

   This excerpt covers repeated delivery. It does not establish behavior under every concurrent workload.

   Back to competency evidence ↑
2. [02]

   Change excerpt

   Illustrative

   Make retries safe at the write boundary.

   event_processor.py · pseudocode diff

   - retry(() => insert(payload))
   + retry(() => insert_once(event.id, payload))
   
   # insert_once uses a unique event ID at persistence.

   What it supports

   The change places duplicate protection where the record is written, so a retry can reuse the same event identity.

   What remains open

   This simplified excerpt illustrates the approach. A reviewer would inspect the full implementation and transaction behavior.

   Back to competency evidence ↑
3. [03]

   Check results

   Illustrative

   Verify the change against the task checks.

   Task check summary · illustrative output

   PASS  duplicate delivery
   PASS  retry after timeout
   PASS  valid event
   PASS  invalid payload
   PASS  empty batch
   PASS  ordered batch
   PASS  persistence failure
   PASS  existing behavior
   
   8 / 8 passed

   What it supports

   The sample distinguishes the reproduced failure from the other task checks, making the scope of verification visible.

   What remains open

   Passing these checks supports the scoped task. It does not establish production readiness or overall job performance.

   Back to competency evidence ↑
4. [04]

   Handoff note

   Illustrative

   Resolve the open question with targeted follow-up.

   Candidate explanation · illustrative excerpt

   “I bounded retries to avoid an endless loop.
   I would monitor retry volume after deployment.”
   
   Not specified: alert threshold or escalation condition.

   What it supports

   The explanation identifies an operational concern, but gives the reviewer insufficient evidence about the monitoring decision.

   What remains open

   Missing detail is a follow-up question. It is not proof that the candidate cannot make the decision.

   Back to competency evidence ↑