US 11,657,069 B1
Dynamic compilation of machine learning models based on hardware configurations
Balakrishnan Narayanaswamy, San Jose, CA (US); Gokul Soundararajan, San Jose, CA (US); Jiayuan Chen, Menlo Park, CA (US); Yannis Papakonstantinou, La Jolla, CA (US); Vuk Ercegovac, Campbell, CA (US); George Constantin Caragea, Redwood City, CA (US); Sriram Krishnamurthy, San Francisco, CA (US); and Nikolaos Koulouris, San Francisco, CA (US)
Assigned to Amazon Technologies, Inc., Seattle, WA (US)
Filed by Amazon Technologies, Inc., Seattle, WA (US)
Filed on Nov. 25, 2020, as Appl. No. 17/105,214.
Int. Cl. G06F 16/28 (2019.01); G06F 16/24 (2019.01); G06F 16/2458 (2019.01); G06N 20/00 (2019.01); G06F 16/2453 (2019.01); G06F 8/41 (2018.01)
CPC G06F 16/283 (2019.01) [G06F 8/447 (2013.01); G06F 16/2465 (2019.01); G06F 16/2471 (2019.01); G06F 16/24535 (2019.01); G06F 16/24542 (2019.01); G06N 20/00 (2019.01)] 20 Claims
OG exemplary drawing
 
1. A system, comprising:
one or more hardware processors and memory configured to implement a database system including a plurality of computing resources, wherein the database system is configured to:
receive a request to create a machine learning model from a first set of data stored in the database system;
provide the first set of data stored in the database system to a machine learning model creation system to train the machine learning model, wherein the machine learning model creation system is implemented using one or more computing resources external to the database system;
obtain an executable version of the machine learning model from the machine learning model creation system according to a hardware configuration of one or more computing resources selected out of the plurality of computing resources of the database system to perform requests to the database system that invoke the machine learning model, wherein information of the hardware configuration is provided from the database system to the machine learning model creation system; and
store the executable version of the machine learning model at the one or more computing resources selected out of the plurality of computing resources of the database system.