GPU Model Lifecycle Management | Dataford Interview Questions

How would you structure a class to manage the lifecycle of a model loaded into GPU memory?

Implement a lifecycle manager that tracks GPU residency with reference counting. The input is a list of operations, each either load, infer, or unload; return the status produced after each operation. No actual GPU calls are required.

Constraints

1 <= len(events) <= 1000
Each event is exactly "load", "infer", or "unload"
The manager starts with no model loaded
The model is released only when its reference count reaches zero

Function Signature

def manage_model_lifecycle(events):

Your question is GPU Model Lifecycle Management. Start with the requirements in the Question tab.

Run and submit as often as you like. When you're ready, talk me through your approach or go straight to the code.

Problem

How would you structure a class to manage the lifecycle of a model loaded into GPU memory?

Constraints

1 <= len(events) <= 1000
Each event is exactly "load", "infer", or "unload"
The manager starts with no model loaded
The model is released only when its reference count reaches zero

Function Signature

def manage_model_lifecycle(events):

Your question is GPU Model Lifecycle Management. Start with the requirements in the Question tab.

Run and submit as often as you like. When you're ready, talk me through your approach or go straight to the code.