PortugalDevelopment of internal Tools: Batch Prediction Service for internal customers Using Flyte, Ray and Cuda
Further improvement of Inference Platform: Speed-up in inference time while reducing costs, increasing throughput 10x. Stack used Ray, FastAPI