AI Prompts

Design a safe, restartable backfill that does not overwhelm production systems or violate SLAs.

Size a streaming platform for peak traffic, ordering, lag, and failure recovery, with formulas you can defend and a load-test plan.

Determine whether a streaming system delivers real end-to-end exactly-once outcomes, or only a framework marketing claim.

Diagnose stragglers, spills, OOMs, and expensive shuffles in large Spark jobs with concrete next checks.

Fix lakehouse small-file amplification while preserving ingestion latency, query scan cost, and safe concurrent writers.

Diagnose why a warehouse query suddenly got slower or more expensive, with likely causes, checks to run, and practical remediations.

Design auditable reconciliation controls for revenue, payments, or ledger pipelines, without silently netting unexplained gaps.

Turn vague data reliability expectations into measurable SLOs, error budgets, and burn-rate alerts consumers can use.

Find sensitive-data exposure and design least-privilege controls across a data platform.

Build and test recovery plans for regional outages, corruption, and credential compromise.

Review an Airflow DAG for idempotency, scheduler pressure, backfills, and safe recovery.

Prove a dbt incremental model handles updates, late data, deletes, and full refreshes, with fixtures that would catch silent drift.