Methods and systems for determining recommendations based on real-time optimization of machine learning models

A method comprising:

• performing, via one or more processors, operations comprising:
• in connection with a request having a processing characteristic, inputting a first feature input derived from the processing characteristic of the request into a first machine learning model to obtain a first output indicating a first domain, wherein the first machine learning model (i) is not constrained to impose a monotonic relationship between outputs and respective domain rule sets for each of a plurality of available domains and (ii) is trained to determine additional processing characteristics based on historical processing data and known processing characteristics, wherein training the first machine learning model comprises applying a linear program solver to the historical processing data to determine optimal data points for unknown processing characteristics based on the known processing characteristics;
• after inputting the first feature input into the first machine learning model not constrained to impose the monotonic relationship, inputting a second feature input derived from the first output into a second machine learning model to obtain a second output indicating the first domain, wherein the second machine learning model (i) is constrained to impose the monotonic relationship between output and respective domain rule sets for each of the plurality of available domains (ii) is trained to determine respective item characteristics, for given requests, for each domain of a plurality of available server domains, and wherein each of the respective item characteristics is based on the first output and respective item rule sets for each of the plurality of available domains;
• and
• based on the second output, executing, with the first domain, a task indicated by the request.

One or more non-transitory, computer-readable media comprising instructions that, when executed by one or more processors, cause operations comprising:

• in connection with a request having a processing characteristic, inputting a first feature input derived from the processing characteristic of the request into a first machine learning model to obtain a first output indicating a first domain, wherein the first machine learning model (i) is not constrained to impose a monotonic relationship between outputs and respective domain rule sets for each of a plurality of available domains and (ii) is trained to determine additional processing characteristics based on historical processing data and known processing characteristics, wherein the first machine learning model is trained by applying a linear program solver to the historical processing data to determine data points for unknown processing characteristics based on the known processing characteristics;
• after inputting the first feature input into the first machine learning model not constrained to impose the monotonic relationship, inputting a second feature input derived from the first output into a second machine learning model to obtain a second output indicating the first domain, wherein the second machine learning model is constrained to impose the monotonic relationship between output and respective domain rule sets for each of the plurality of available domains;
• and
• based on the second output, executing a task indicated by the request with the first domain.