Workload distribution and execution platform for tasks such as artificial intelligence tasks

One or more computer memories collectively containing a client status data structure relating to a plurality of software modules, the data structure comprising:

• for each of a plurality of client computer systems in a local network, private network, or corporate network, each configured for use by a local user:
• a level of availability of the client computer system to perform external requests, based at least in part on a level at which a local user of the client computer system has recently used the client computer system, and
• a level of suitedness of the client computer system to perform external requests, based at least in part on:
• for each of the plurality of software modules, a level of availability of the software module on the client computer system;
• and
• a predicted level of performance of the client computer system in executing the software module, wherein the predicted level of performance is determined by:
• obtaining a base performance metric for the client computer system by causing the client computer system to execute an operation relevant to executing the computing task using the identified machine learning model;
• such that the contents of the data structure are usable to, for a request that an identified software module be applied to particular input data, select a client computer system among the plurality of client computer systems to which to route the request.

One or more computer memories collectively having contents configured to cause client software module executing on a client computer system in a local network, private network, or corporate network to perform a method, the method comprising:

• periodically reporting operating data for the client computer system, comprising:
• first data reflecting a level of availability of the client computer system to perform external requests, based at least in part on a level at which a local user of the client computer system has recently used the client computer system,
• second data reflecting, for each of a plurality of software modules, a level of availability of the software module on the client computer system, and/or
• third data reflecting, for each of the plurality of software modules, a predicted level of performance of the client computer system in executing the software module, wherein the predicted level of performance is determined by:
• obtaining a base performance metric for the client computer system by causing the client computer system to execute an operation relevant to executing the computing task using the identified machine learning model;
• receiving a request delegated to the client computer system in response to selection of the client computer system from among a plurality of client computer systems at least partly on the basis of the operating data reported for the client computer system, the delegated request comprising:
• information identifying a software module, and
• input data upon which the identified software module is to be executed;
• in response to receiving the delegated request:
• causing the identified software module to be executed upon the input data on the client computer system to produce execution results;
• and
• responding to the request with the execution results.

A secure computer network comprising:

• a plurality of client nodes in a local network, private network, or corporate network, each assigned for use by at least one local user, each of the client nodes configured to perform a first method, the first method comprising:
• periodically reporting operating data for the client node, comprising:
• first data reflecting a level of availability of the client node to perform external requests, based at least in part on a level at which a local user of the client node has recently used the client computer system,
• second data reflecting, for each of a plurality of software modules, a level of availability of the software module on the client node, and/or
• third data reflecting, for each of the plurality of software modules, a predicted level of performance of the client node in executing the software module, wherein the predicted level of performance is determined by:
• obtaining a base performance metric for the client computer system by causing the client computer system to execute an operation relevant to executing the computing task using the identified machine learning model,
• a routing node configured to perform a second method, the second method comprising:
• receiving operating data from client nodes of the plurality of client nodes;
• using the received operating data to update a client status resource that indicates, for each of the plurality of client nodes:
• a level of availability of the client node to perform external requests, and
• a level of suitedness of the client node to perform external requests;
• receiving from one or more requestor nodes a series of requests, each request specifying a computing task to be performed and comprising:
• information identifying a software module, and
• input data upon which the identified software module is to be executed;
• in response to receiving each request:
• accessing the client status resource;
• using the contents of the accessed client status resource to select one of the plurality of client nodes;
• causing the request to be delegated to the selected client node, in such a way that execution results produced by executing the software module identified the by the request upon the input data identified by the request is conveyed to the requestor node;
• the first method further comprising:
• for each request delegated to the client node:
• causing the identified software module to be executed upon the input data on the client node to produce execution results;
• and
• responding to the request with the execution results.