US 12,481,539 B2
Workload optimization through contextual bandits
Akshay Ravindran, Edmonton (CA)
Assigned to Intuit Inc., Mountain View, CA (US)
Filed by Intuit, Inc., Mountain View, CA (US)
Filed on Mar. 20, 2024, as Appl. No. 18/610,793.
Application 18/610,793 is a continuation of application No. 18/215,756, filed on Jun. 28, 2023, granted, now 11,983,574.
Prior Publication US 2025/0004852 A1, Jan. 2, 2025
Int. Cl. G06F 9/50 (2006.01)
CPC G06F 9/505 (2013.01) [G06F 9/5033 (2013.01); G06F 9/5044 (2013.01)] 20 Claims
OG exemplary drawing
 
1. A computer-implemented method, comprising:
receiving workload performance data associated with a workload processing on a remote processing system using a first processing configuration comprising a first set of resource requirements;
determining, based on the workload performance data and the first processing configuration, at least one resource requirement of the first processing configuration is insufficient for processing the workload on the remote processing system;
determining one or more workload requirements associated with the workload based on the workload performance data and the first processing configuration;
setting at least one resource requirement constraint associated with the one or more workload requirements based on the workload performance data and the first processing configuration;
providing, to a machine learning model, the workload, and the one or more workload requirements, and the at least one resource requirement constraint, wherein the machine learning model is trained to reduce processing resource usage by the remote processing system while meeting the one or more workload requirements;
receiving, from the machine learning model a second processing configuration comprising a second set of resource requirements for processing the workload on the remote processing system and satisfying the at least one resource requirement constraint; and
transmitting, to the remote processing system, an indication of one or more resources to be provisioned at the remote processing system based on the second processing configuration thereby causing execution of the workload at the remote processing system.