Discuss deadlines, feedback, and deep dives

# How to Answer - Use STAR: Situation (context), Task (your responsibility), Action (specific steps), Result (measurable outcome). Add your Reasoning and Trade-offs. - Quantify: latency (p95/p99), error rates, cost, on-call load, customer impact, cycle time. - Verify: show how you knew it worked (dashboards, tests, canary, A/B, customer metrics). 1) Tight Deadline (Sample Answer) - Situation: Security flagged a critical zero‑day in our auth library; we had 72 hours to patch 3 services without downtime. - Task: Lead the patch, testing, and rollout across services. - Action: - Planned with MoSCoW: Must (upgrade library, regression tests, canary), Should (rollout automation), Could (refactor auth wrappers), Won’t (non-blocking refactors now). - Timeboxed: 4 hours test harness, 8 hours patch + unit/integration tests, 4 hours canary + observability, buffer for rollback. - Implemented feature flag + 10%→50%→100% canary, created rollback plan, and ran synthetic login tests in CI. - Trade-offs: Deferred wrapper refactor and doc polish to meet the SLA; focused on minimal, reversible changes. - Result: Shipped in 36 hours, zero downtime, no auth errors, p99 unchanged, passed security scan on first pass. Why this works: Shows planning, prioritization, explicit trade-offs, and measurable success with guardrails. 2) Missed Deadline (Sample Answer) - Situation: Committed to deliver an event consumer by Friday for a launch dependency. - Task: Integrate with an upstream event schema. - Action: - On Wednesday the upstream schema changed; pipeline failed. I immediately notified PM/partner team with impact options: 1) ship partial read-only logging by Friday, 2) slip by 2 business days to ship fully with dual-schema support. - Agreed to slip with daily status. Mitigation: delivered a stub logger so partner could validate event volume; wrote consumer-driven contract tests; added dual-version parsing. - Created a fallback toggle and documented risks in an issue. - Result: Shipped Tuesday with backward compatibility, zero data loss, 100% green e2e. Stakeholders unblocked testing. Postmortem led to a schema-freeze checklist and CDC tests in CI. - Learning: Always secure schema freeze, add consumer-driven contracts, and buffer for external dependencies; escalate early with options. 3) Negative Feedback (Sample Answer) - Situation: In design reviews, feedback was that I didn’t present trade-offs clearly and got defensive in comments. - Task: Improve design communication and collaboration. - Action: - Adopted an ADR template: Problem, Options (A/B/C), Pros/Cons, Decision, Risks, Rollback. - Summarized reviewer feedback in the doc, responded with evidence, and scheduled a 15‑minute sync for blocking items. - Implemented a 24‑hour “cool-off” before replying to contentious reviews. - Result: Next 3 designs approved in a single review round; review cycle time fell ~30%; peer survey noted clearer trade-off articulation. - Follow-up: Asked the same reviewers 6 weeks later for re-evaluation; kept the ADR template as a team standard. 4) Beyond Requirements (Sample Answer) - Situation: Requirement was to build a new job scheduler endpoint. - Task: Deliver API and basic monitoring. - Action: - Beyond scope, I added SLOs (99.9% job start within 2 min), exported Prometheus metrics, Grafana dashboards, and an auto‑rollback on elevated failure rate. - Implemented a load test in CI and a runbook with one‑click rollback. - Result: MTTR dropped from ~20 min to ~4 min; paging incidents reduced 40% over 2 months; caught a memory leak pre‑launch, avoiding an outage. - Why: Operational excellence reduces long-term toil and risk; minimal extra effort at design time pays off repeatedly. 5) Dive Deep (Sample Answer) - Situation: Checkout p95 latency spiked from ~220 ms to ~900 ms after a release. - Task: Find root cause and fix without blocking orders. - Action: - Hypotheses: N+1 DB queries, missing index, thread pool saturation, GC pauses, third‑party latency. - Data: Compared traces (before/after), DB slow query log, JVM GC logs, thread dumps, dependency health checks. - Findings: A new ORM call caused an N+1 (1 order → 17 item queries). Verified by query count/time in traces. - Fix: Rewrote to a single JOIN with proper index, added read‑through cache, and regression test counting queries. - Verification: Canary at 10% → 100%, dashboards stable, A/B showed p95 back to ~230 ms, error rate unchanged. - Result: Sustained p95 ~230 ms, 0 rollback needed, added a tracing alert for query explosions to prevent recurrence. Tips and Pitfalls - Pick distinct examples; show ownership, communication, and verification. - Always provide options and rationale; show what you did not do and why. - Close the loop: how you ensured the fix or behavior change stuck (tests, alerts, check-ins, docs). 6) Reverse Questions (Ask the Hiring Manager) - Expectations: "In the first 90 days, what outcomes define success for this role, and how do you measure engineering impact (e.g., delivery, reliability, customer metrics)?" - Team Practices: "What are the team’s norms around code review SLAs, testing, release cadence, and on-call? How do you balance speed vs. reliability (SLOs/error budgets)?"