Interview Guides

You are deploying a perception model that must run on a vehicle in real time. You need to balance prediction quality with tight onboard compute and latency limits.
What techniques would you use for model optimization to ensure real-time performance on a vehicle?