US 12,481,534 B2
Dynamically selecting artificial intelligence models and hardware environments to execute tasks
Ashok Pancily Poothiyot, San Francisco, CA (US); Ali Zafar, Fremont, CA (US); Anthony Penta, Bellevue, WA (US); Stephen Voorhees, Alamo, CA (US); Tim Gasser, Austin, TX (US); Tsung-Hsiang Chang, Bellevue, WA (US); and Geoff Hulten, Lynnwood, WA (US)
Assigned to Dropbox, Inc., San Francisco, CA (US)
Filed by Dropbox, Inc., San Francisco, CA (US)
Filed on Jun. 3, 2024, as Appl. No. 18/732,297.
Claims priority of provisional application 63/623,662, filed on Jan. 22, 2024.
Prior Publication US 2025/0238264 A1, Jul. 24, 2025
Int. Cl. G06F 9/50 (2006.01); G06F 9/48 (2006.01); G06F 11/14 (2006.01); G06F 11/20 (2006.01); H04L 41/16 (2022.01); H04L 43/08 (2022.01); H04L 43/0894 (2022.01)
CPC G06F 9/5027 (2013.01) [G06F 9/48 (2013.01); G06F 9/4806 (2013.01); G06F 9/4843 (2013.01); G06F 9/4881 (2013.01); G06F 9/50 (2013.01); G06F 9/5005 (2013.01); G06F 9/5055 (2013.01); G06F 11/1476 (2013.01); G06F 11/2023 (2013.01); G06F 11/2038 (2013.01); H04L 41/16 (2013.01); H04L 43/0894 (2013.01); G06F 2209/501 (2013.01); G06F 2209/503 (2013.01); G06F 2209/509 (2013.01)] 20 Claims
OG exemplary drawing
 
1. A computer-implemented method comprising:
receiving, from a device connected by a network to a content management system, workload data requesting execution of a task using a machine-learning model;
extracting, from the workload data, workload features defining estimated computational requirements for executing the task;
determining task routing metrics indicating an availability status for a plurality of machine-learning models hosted in respective network environments;
generating a software domain analysis by analyzing the plurality of machine-learning models to identify a plurality of common features and a plurality of variable features for machine-learning models of the plurality of machine-learning models;
generating an optimization metric for each machine-learning model of the plurality of machine-learning models based on a combination of the workload features for the task, the software domain analysis, and the task routing metrics; and
selecting, utilizing a model selection machine-learning model, a designated machine-learning model from the plurality of machine-learning models for executing the task based on a given optimization metric for the designated machine-learning model.