Just finished the full loop at Citadel for a senior SWE role in the securities technology group. Wanted to put down specifics because the existing posts are vague.
The system design round at Citadel is not your typical LeetCode-adjacent design round. They care a lot about latency, throughput, and the reliability of distributed systems. Think: build a low-latency order routing system, or design a market data feed aggregator that handles 500k events/sec. They're not checking if you know the buzzwords. They want to see how you reason under constraint.
What got picked apart in my session: How I'd handle backpressure when the feed spikes My approach to idempotency on the trade confirmation side How I'd partition a time-series store for tick data
The interviewer was an L6 engineer and pushed hard on every decision. Not mean, just relentless. This is not the place to say "we could use Kafka" and stop there. You need to go two or three layers deep on every component.
What helped me: I spent three weeks before the loop working through distributed systems material specifically through a financial-systems lens, not just standard FAANG prep. Order matching, event sourcing, CDC patterns. If you haven't thought about what happens to your system when a network partition hits mid-trade, you're not ready.
Compensation wasn't discussed in the design round obviously, but for senior roles in NYC 2026 the total comp I've seen benchmarked around $350-450k depending on the group and your negotiation. The trading technology orgs typically pay above the securit side.
Round itself: 55-60 minutes, first 5 minutes on your background, then they dive straight in. No "let me explain the problem" delay. The prompt is intentionally underspecified and you're expected to ask clarifying questions immediately.
Happy to answer specific questions about format.