Human-in-the-Loop Control of Objective Drift in LLM-Assisted Computer Science Education

Create a library for 'objective drift' detection in AI programming assistants. Developers need a way to detect when an LLM deviates from the user's explicit task requirements.

Suggested repo: drift-watch

"Is your coding assistant drifting off course? Catch the deviation before it becomes a bug."

Estimated effort: 30h