ServiceNow · Primly Community

ServiceNow senior / L5 system design interview, what to expect

sre_sol · 4 replies

Just cleared my ServiceNow final round for a Senior SWE role (what they level as L5 equivalent internally, they call it 'Staff Software Engineer' on the JD but it's senior comp territory). Sharing the system design round because that's where I spent 75% of my prep.

The prompt I got: design a notification delivery system that can fan out to millions of users across email, SMS, push, and in-app channels with configurable retry logic and SLA tracking. Classic enterprise workflow territory for a company whose bread and butter is ITSM.

What they actually care about: Reliability and failure handling above all. ServiceNow runs enterprise incident workflows. Downtime or dropped notifications costs their customers real money. My interviewer lit up when I talked through dead-letter queues and idempotency keys. Horizontal scaling. I sketched out a message queue layer (Kafka or Pulsar, I mentioned both), sharded worker pools, and a rate-limiter per channel. They pushed back with "what happens when the SMS gateway goes down mid-burst" which is exactly the kind of prod scenario they care about. Operational observability. They asked about metrics, tracing, alerting almost unprompted. I walked through latency histograms per channel, failure rate SLOs, and a retry dashboard. That landed well.

What did NOT seem to matter much: Cutting-edge distributed consensus stuff, microservices-at-Google-scale theory. They're an enterprise SaaS shop. Boring reliable beats clever novel.

Timing: 45 minutes. I spent maybe 8 minutes on high-level design, 25 on deep dives they directed, 10 on trade-offs and wrap-up. No coding in this round.

Prep that helped: Designing Data-Intensive Applications for the queue/durability sections. The system design questions in distributed systems channels here. Also brushing up on ServiceNow's own Now Platform architecture (public blog posts) gave me useful context to drop into my answers.

Happy to answer specifics. My loop was 4 rounds total: 1 coding, 1 system design, 1 behavioral, 1 hiring manager chat.

4 replies

pivot_pat

The fan-out notification design is such a classic ServiceNow prompt. I got basically the same thing two years ago. Did they ask you how you'd handle priority queues? Like urgent P1 incident alerts vs. routine digest emails?

remote_swe_42

Yes, exactly that. I separated them into priority tiers with different consumer groups. P1 alerts hit a dedicated pool of workers with no backpressure, lower priority traffic goes through the standard throttled path. They seemed to like having explicit SLA contracts per tier.

jp_newgrad

Is this same prompt / format for a mid-level SWE role or do they adjust complexity for different levels?

remote_swe_42

From what I heard, mid-level gets a similar format but they expect less unprompted depth on the ops/observability side. More about correctness of the design, less about production edge cases. Don't quote me on that.