THE AI infra BLOG · FREE & OPEN
How to land a AI infrastructure engineer offer.
Deep, no-fluff articles on the AI infra and Applied AI interview: company loops, the rounds that actually decide it, prep plans, and the AI skills that get offers. Researched from real loops, with a question bank to practice on.
"Your All-Reduce Got Slow": How to Answer the Most Common Distributed Training Debug Question
A training run that was doing 40 percent MFU is now doing 22, and the interviewer wants your first three moves. Here is the narrowing sequence that separates a strong answer from a list of guesses, worked with real numbers.
Jonas Weber · Sep 6, 2026 · 11 min read
AI Infra Coding Interviews Are Not LeetCode: What They Ask Instead
Rate limiters, resumable iterators, ledgers, worker pools and schedulers, usually progressive, usually with a production follow-up. Here is what infrastructure coding rounds actually test and how to practise for them.
Morgan Foster · Sep 6, 2026 · 10 min read
Anthropic Infrastructure and Performance Engineering Interviews: A Preparation Guide
Anthropic publishes its own performance take-home, which makes this loop unusually knowable. Here is the reported process, what the simulated-accelerator exercise is really testing, and why the values round rejects more candidates than the coding rounds.
Deepa Venkatesh · Sep 6, 2026 · 10 min read
CoreWeave Interview Guide: Go Concurrency, Kubernetes and GPU Scheduling
CoreWeave's engineering loop is Go-first, concurrency-first and Kubernetes-native. Two first-hand 2025 debriefs plus Glassdoor reports agree on the shape. Here is what each round tests and what the rejection feedback keeps saying.
Devin Porter · Sep 6, 2026 · 9 min read
KV Cache Interview Questions: The Arithmetic That Decides Your Serving Answer
The KV cache, not the model weights, is what limits how many users fit on a GPU. Here are the questions interviewers ask about it, the formula they expect you to derive, and the six ways to make the number smaller.
Maya Castillo · Sep 6, 2026 · 10 min read
Napkin Math for AI Infrastructure Interviews: The Numbers You Must Be Able to Derive
Almost every AI infra round has arithmetic hiding in it. Here are the six chains worth memorising, the constants you need, and the derivations, so you can size a KV cache, a training run or a serving fleet without a calculator.
Brandon Sullivan · Sep 6, 2026 · 11 min read
NVIDIA AI Infrastructure Engineer Interview: Rounds, Questions and How to Prepare
What NVIDIA's AI infrastructure and kernel loops actually test: C++ fundamentals, domain depth on the specific team, and a systems architecture round. Reported rounds, the questions that recur, and a preparation order.
Ruiqi Zhou · Sep 6, 2026 · 9 min read
The GPU Cluster Design Round: How to Size a Training Cluster Out Loud
Design rounds for AI infrastructure are sizing exercises, not box-drawing exercises. Here is the arithmetic chain that carries the round, worked end to end for a 70B training run, plus the failure modes interviewers wait for you to name.
Owen Hartley · Sep 6, 2026 · 11 min read
The Kernel Take-Home: How to Approach a GPU Optimisation Exercise Under a Deadline
Kernel and performance take-homes give you a slow program and a few hours. The candidates who do well measure first, name the ceiling, and change the term that binds. Here is the working order, and the traps that eat the clock.
Ruiqi Zhou · Sep 6, 2026 · 10 min read
The OpenAI Infrastructure Engineer Interview: What the Loop Actually Tests
OpenAI's infrastructure loop is coding-heavy and design-heavy, and neither round looks like the LeetCode you practised. Here is the reported shape, the question patterns candidates keep describing, and what to prepare in what order.
Maya Castillo · Sep 6, 2026 · 10 min read
Done reading? Start practicing.
The must-know AI infra questions →