The process had multiple stages: first a coding exercise - not LeetCode, an actual problem related to what you'd do day to day at Datadog. Then an Agent Design round covering pipeline architecture, Linux internals, and fault tolerance. An AI Coding round where they actually watch how you work and think, not just whether you get the right answer. And a Hiring Manager round that went deep into a past project and behavioral stuff.
What really stood out is how well everything was put together. Every stage felt relevant to the actual job. The recruiters were responsive, the coordinator kept things moving, and the prep material for each stage was specific and actually matched what came up in the interviews. The interviewers clearly prepared too - it felt like a real conversation, not someone reading off a rubric. No trick questions, no algorithm puzzles pulled from nowhere.
If you're preparing: know your Linux fundamentals for real, not just theory. Pick a project you owned end to end and be ready to talk about what went wrong, not just what went right. For the AI Coding round, think out loud and work in small steps.
