Vanguard · Primly Community

Vanguard senior / L5 system design interview, what to expect (went through it last quarter)

remote_swe_42 · 5 replies

Finished a Vanguard loop for a Senior SWE role in Malvern a few months ago. Going to share what I actually saw in the system design round because there's almost nothing useful about this specific company online.

Vanguard calls the level roughly equivalent to L5 at Google or E5 at Meta. The system design interview was 60 minutes, one interviewer, a shared virtual whiteboard.

The prompt they gave me: design a system that tracks real-time NAV (net asset value) updates for a large mutual fund portfolio and surfaces alerts when a fund deviates from its benchmark by more than some threshold.

What mattered: They actually cared about data freshness tradeoffs. I went deep on event streaming (Kafka-style) vs. polling, and the interviewer engaged hard on exactly that. Latency vs. correctness was a real theme. They pushed me on what happens if you process a stale price. Financial data has regulatory implications they take seriously. They wanted to know how you'd handle scale without going nuts. This isn't a consumer-facing TikTok-tier product. Throughput is in the tens of thousands of fund updates per day, not billions of requests. Calibrate accordingly. Storage: they wanted me to think about time-series data, historical snapshots, audit logs. Compliance is a big deal at Vanguard.

What they were less focused on: sharding schemes for petabyte scale, CDN strategies, anything that felt pure FAANG. They care about correctness, auditability, and reliability. Think boring finance infra, not Google Maps.

The interviewer was a principal-level engineer. Asked good follow-up questions, not gotcha style. They mentioned they're building on AWS internally and modernizing off some legacy on-prem stuff, so cloud migration context came up naturally.

Overall difficulty: medium-hard. Easier than FAANG system design if you understand financial data concepts. Harder if you've only studied "design Twitter."

I declined the offer for comp reasons (more on that in another thread) but the loop itself was one of the more technically honest processes I've been through.

5 replies

ae_andre

The NAV prompt is interesting. Did they care much about the difference between end-of-day NAV (standard mutual fund calc) vs. intraday estimated NAV for ETFs? That's a meaningful architectural fork. Intraday means near-real-time market feed integration, end-of-day is basically a batch job.

sdr_sky

Yeah they went there. They framed it as "assume mutual funds for now but design for future ETF support." So you had to at least acknowledge the intraday feed problem and not paint yourself into a corner. The system design had a natural extension point conversation.

marketer_mei

The boring finance infra point is real. I've interviewed at a few asset managers and the calibration is consistently "we need it correct, auditable, and recoverable." Nobody cares if your p99 latency is 12ms vs 8ms. They care if a bad trade went through because your system had a race condition.

content_cole

Did they ask any frontend system design at all? I'm a frontend eng eyeing Vanguard and wondering if the loop is identical across specializations or if they fork the design round.

pivot_pat

My loop was backend-focused so I can't speak to frontend. Worth asking the recruiter specifically what the design round covers for your track. My recruiter was pretty responsive about that kind of meta-question.