Basic Load Balancing for Inference

Describe how you would implement a basic load-balancing logic for distributed inference requests.

Write a function that takes the current load on each worker and a list of request costs, then returns the worker index chosen for each request. For each request, assign it to the worker with the smallest current total load, breaking ties by smaller worker index, and then add that request's cost to that worker.

Function signature

def balance_requests(worker_loads, requests):

Input

• worker_loads: list of integers, initial load per worker
• requests: list of integers, cost of each incoming request

Output

• Return a list of integers, where each value is the chosen worker index for the corresponding request.