Using AI to write summarize this post ofcourse.
Role: Software Engineer E4, Full Stack, Seattle, hybrid
Timeline: About 4-5 weeks from recruiter call to the end of the onsite
Format: 6 conversations: 1 screen, 1 HM chat, 4 onsite rounds over 2 days
1. Frontend coding screen (HackerRank, ~45 min)
- Fetch data from a REST API and render it as a responsive card grid.
- Handle loading and error states, and format currency correctly.
- The follow-up asks how you'd improve the code: component reuse, caching, filters, strong typing.
- Prep: GreatFrontEnd "fetch and render list" questions, or build a small React/vanilla JS list with async states in under 30 min.
2. Hiring manager chat (~45 min)
- Team overview, role expectations (about 80% backend, 20% frontend), and how the team works with AI agents.
- Not a coding round, but come with good questions. They care a lot about AI fluency and ownership.
3. Onsite: System design (~60 min, no coding)
- Design a payments-style flow on top of a third-party payment processor during a short, high-traffic event.
- Key topics: idempotency keys, avoiding double charges, retries with exponential backoff, dead-letter queues, a queue to absorb traffic spikes, back-of-envelope QPS math, API and data model design, and SQL vs NoSQL tradeoffs.
- Prep:
- Hello Interview's "Design a Payment System."
- "Design Ticketmaster," for spiky traffic.
- Stripe's blog post on idempotent requests.
4. Onsite: Engineering values + AI usage (with the HM, ~60 min)
- Behavioral, built around DoorDash values: ownership, disagreeing with another team, acting on feedback, career goals.
- Expect a deep dive into a project you built from scratch.
- They also ask how you use AI in your day-to-day work, from design to code.
- Prep: Have 2-3 strong STAR stories (Situation, Task, Action, Result) ready. Read DoorDash's mission and values page.
5. Onsite: CodeCraft (Java/Kotlin/Python/Go, ~50 min)
- Practical OOP/API design, not LeetCode.
- Build a backend endpoint that pulls data from several internal services into one response.
- Key topics: critical vs non-critical dependencies, returning a partial response when a non-critical service fails, a clean typed data model, and making it testable.
- The follow-up covers resilience: retries, backoff with jitter, and circuit breakers.
- Prep: Practice "API gateway / BFF fan-out aggregation" problems. Read the Resilience4j circuit breaker docs.
6. Onsite: AI CodeCraft (~60 min, new pilot round)
- Build a frontend feature (live dasher tracking using sse) step by step on top of a starter Node backend, using a real AI coding agent (Claude Code, Cursor, Codex). A ChatGPT chat window alone is not allowed. Full screen share.
- The task had milestones, and the later ones involved real-time updates.
- They judge prompt quality, why you pick each model, engineering judgment, how you validate AI output (unit tests, Playwright), and how you'd structure and review PRs.
- Prep: Build a small React dashboard with live WebSocket updates in Claude Code or Cursor, timed to 1 hour. Practice saying out loud why you accept or reject the AI's code.
Overall
- No LeetCode-style DSA. Everything was practical and close to the real job.
- Interviewers were friendly and conversational. The recruiter was clear about the format and timeline.
- Biggest tip: in the AI round, treat it like a normal workday task and explain your thinking out loud.
Comments (2)