编程开发

Production task queue (BullMQ/Sidekiq/Celery/Temporal-style): retries, priorities, scheduling, idempotency, and visibility.

Product analytics ingestion: schema, SDK, batching, idempotency, warehouse load.

Robust bulk data in/out: streaming, validation, error reports, huge files, encoding issues, and idempotent upserts.

Transactional + marketing notification system across channels with deliverability, templates, preferences, and suppression.

Implements live collaboration/chat/notifications with presence, reconnection, message ordering, and backpressure.

Reliable billing: idempotent webhooks, proration, dunning, plan changes, and reconciliation — the area where silent bugs cost real money.

Designs and codes a maintainable permission system: role hierarchy, resource-level checks, policy evaluation, and tests — preventing authorization sprawl.

End-to-end SaaS feature (billing-aware, tenant-scoped, audit-logged) across stack. Use for subscription-product work where isolation/compliance matters.

Defines release management: branching, versioning, environments, approval, and deployment patterns matching team risk.

Builds automated + manual WCAG 2.2 AA testing into component libraries and CI.

Production-grade DAG design: idempotency, partitioning, backfills, SLA sensors, and data-quality gates.

External black-box probes for availability and key transactions (login, checkout) across regions, with meaningful alerting.

Speeds up slow pipelines via caching, affected-only builds, test splitting, and profiling where time actually goes.

Safe rollout system: flag lifecycle, targeting, kill switches, experimentation guardrails, and cleanup.

Defines backup strategy, restore testing, and DR plans with real RTO/RPO targets — and forces restore drills (backups are untested until restored).

Writes actionable runbooks for common alerts: triage decision tree, commands, mitigation before diagnosis, escalation thresholds.

Turns a schema change into a safe zero-downtime runbook: expand/contract steps, locks analysis, verification, rollback.

Reviews/writes production-grade K8s config: probes, resources, securityContext, autoscaling, PDBs, and rollout safety.

Produces small, secure, reproducible images: layer caching, multi-stage builds, non-root, SBOM-ready, with run/diagnose commands.

Creates realistic load tests with proper workload models, thresholds, and interpretation — not just "hit the endpoint 10k times."

Designs controlled failure injection to verify retry/fallback/deadline/circuit breaker behavior actually works under load.

Finds invariants and writes property-based tests (Hypothesis/fast-check) that discover edge cases humans miss.

Produces a complete test strategy for a service: pyramid mapping, environments, data, ownership, and release gates — aligned to risk rather than coverage theater.

there 2 simpler alternatives worth comparing? Return findings as BLOCKING QUESTIONS /…