CentML

At CentML our mission is to enable everyone (who might not have any background in system and hardware) to achieve low latencies, high throughputs, and good hardware utilization, thereby reducing the cost, of their inference services and training workloads without affecting model quality.