Design a Simulation Job Scheduler
A job scheduler takes a pile of work and a pile of machines and decides what runs where. Every job here says what it needs: CPU cores, memory, and sometimes GPUs, the graphics processors that render camera and lidar frames. The scheduler finds a machine with that much free, starts the job, watches it, and frees the space when it ends. The picture to keep in mind is a regression suite. Someone merges a change to the autonomy software, CI (the continuous integration system that builds and tests every change) kicks off a build, and the build fans out into thousands of scenario runs. Most of them are CPU-only simulations of physics and planning. Some render sensors and need a GPU. A few are big enough to need several GPU machines at once. All of them want to finish before the engineer loses patience.
Candidates report this prompt in Applied Intuition onsite design rounds in a few forms. The hardware-matching version above is the most specific. A broader one asks for a multi-tenant job scheduler for backend services, with delayed, recurring and dependent jobs, priorities with fair share, retries, a dead-letter queue and idempotent submits. Sometimes it's just a design round with a scheduler focus. One core design covers all three, and the broad version's extras are in the follow-ups.
Applied IntuitionPremium lesson · System design round
The Applied Intuition design, worked end to end.
- The API and data modelSystem design · Frequently asked
- The design, drawn out1 diagram
- 5 deep divesWith the tradeoffs
$49 a year · all 107 companies · instant access
20,000+ candidate reports · guide updated
Applied Intuition interview guideKnow what Applied Intuition asks before you walk in.
10 coding · 4 system design · 15 behavioral questions
Get their questions first.
InterviewLegend is independent and not affiliated with or endorsed by any company it covers.