Went through a senior SWE loop at LM's Space division (Sunnyvale) in early 2026. The system design round stood out enough that I want to write it up separately.
First: LM doesn't call their levels L5. But the senior band is roughly comparable to a senior IC at a big tech company. They care about scope, not label.
What the system design round is NOT Not a Google-style 'design Twitter' or 'design a distributed key-value store' exercise. They're not testing whether you know consistent hashing or Dynamo paper. At least not in the space / avionics context.
What it actually is The prompt I got was along the lines of: you're designing a telemetry data ingestion system for a satellite. Constraints are strict: limited bandwidth uplink, need to prioritize certain sensor streams, data must be logged with no loss. How do you approach this?
They want to hear: how you handle tradeoffs under hard real-world constraints, your understanding of fault tolerance (not just "add a retry" -- actual recovery semantics), and whether you ask clarifying questions or just start whiteboarding.
I spent the first 10 minutes clarifying. They seemed to appreciate this. 'What's the acceptable message loss rate?' 'What's the downstream consumer?' That kind of thing.
Format Virtual. They gave me a whiteboard tool (Miro). About 50 minutes. Two interviewers: a staff engineer and a principal. Principal mostly listened. Staff engineer asked follow-ups throughout.
What they asked at the end Almost 15 minutes of 'what would you change if you had 6 months?' and 'where are the failure modes in your design?' This is the evaluation of your ability to critique your own work. Don't be defensive.
The level calibration question They also asked whether I'd owned technical direction on a project. Not just contributed. This felt like a senior vs staff leveling check. For a senior role, one example of leading a technical effort was enough.
Honestly it was one of the more intellectually interesting design rounds I've done. Different flavor than big tech but not easier.