How to pass the Nscale interview, the UK AI neocloud
8 min read The Nscale interview: how the GPU-infra engineering loop is structured, what each round screens for, and how to prepare for it. Read article
8 min read The Nscale interview: how the GPU-infra engineering loop is structured, what each round screens for, and how to prepare for it. Read article
9 min read What Emerald AI screens for across its engineering rounds, how the loop is structured, and how to prepare for the grid-plus-compute questions. Read article
8 min read What Fluidstack’s engineering interview loop looks like in 2026, the systems questions they ask, and how to read the comp ranges. Read article
9 min read A walkthrough of the Kafka questions interviewers actually ask, covering partitions, consumer groups, rebalancing, and delivery guarantees. Read article
7 min read The math behind hash rings, why virtual nodes exist, and what a designer probes the moment you reach for consistent hashing. Read article
7 min read A walkthrough of the rate limiter system design question, from scoping the limit to the distributed counter problem that trips most candidates up. Read article
7 min read A walkthrough of the web crawler design question, from the URL frontier and politeness to deduplication and the follow-ups that trip candidates up. Read article
7 min read The back-of-envelope math, connection routing, and store-and-forward tricks behind moving a few hundred bytes to the right phone among two billion. Read article
6 min read The design choices behind a URL shortener that interviewers actually probe, from short-code generation to caching and the 301-versus-302 redirect. Read article
6 min read A breakdown of the Databricks engineering loop, from the CodeSignal screen through the concurrency round to the distributed-systems design most candidates underprepare for. Read article
9 min read Plain definitions for the system design vocabulary that shows up in interviews, from p99 and quorum to backpressure and idempotency. Read article
8 min read A single-server rate limiter is a hash map and a timer; the real interview starts when you add a second server and need one shared, race-free counter. Read article
6 min read How Materialize runs its engineering loop, and the streaming-systems and correctness questions to prepare for. Read article
What's missing, wrong, or worth adding? One line helps.
Sent — thank you. We read every one.