Automated multi-stage computer code generation

A method, comprising:

• causing, by one or more processors, responsive to inputting one or more prompts including a description of a test problem for evaluating skills of a user to write computer code, into one or more ML models, the one or more ML models to generate;
• (i) an ML generated candidate solution for the test problem, the ML generated candidate solution comprising an ML generated computer code;
• and
• (ii) an ML generated JavaScript Object Notation (JSON) test case object that includes at least one input value and at least one output value to validate the ML generated candidate solution for the test problem;
• passing, by the one or more processors, to a function executed in an execution environment, the at least one input value of the ML generated test case object using the ML generated candidate solution;
• validating, by the one or more processors, by executing the ML generated test case object against the function in the execution environment, the ML generated candidate solution by passing the at least one input value of the ML generated test case object to the function as an argument and comparing the output of the function utilizing the ML generated candidate solution with the at least one output value of the ML generated test case object;
• and
• storing, by the one or more processors, in one or more data structures, the description of the problem and the ML generated candidate solution responsive to validating that the at least one output value of the ML generated test case object matches the output of the function.

A non-transitory computer-readable media having processor readable instructions, such that, when executed, causes at least one processor to:

• cause, responsive to inputting one or more prompts including a description of a test problem for evaluating skills of a user to write computer code into one or more ML models, the one or more ML models to generate:
• (i) an ML generated candidate solution for the test problem, the ML generated candidate solution comprising an ML generated computer code;
• and
• (ii) an ML generated JavaScript Object Notation (JSON) test case object that includes at least one input value and at least one output value to validate the ML generated candidate solution for the test problem;
• pass, to a function executed in an execution environment, the at least one input value of the ML generated test case object using the ML generated candidate solution;
• validate, by executing the ML generated test case object against the function in the execution environment, the ML generated candidate solution by passing the at least one input value of the ML generated test case object to the function as an argument and comparing the output of the function utilizing the ML generated candidate solution with the at least one output value of the ML generated test case object;
• and
• store, in one or more data structures, the description of the problem and the ML generated candidate solution responsive to the validation that the at least one output value of the ML generated test case object matches the output of the function.