Skip to content

Enabling Efficient Machine Learning Model Serving by Minimizing Network Overheads with gRPC

At DoorDash, we confronted similar issues in our gRPC-based model serving setup for search ranking with network overheads taking up to 50 percent of service response time.