# Explain an AI-Assisted Project Describe a project where AI was used, the problem it addressed, your contribution, how outputs were checked, and where you deliberately did not rely on the model. ### Constraints & Assumptions - Use a real project while omitting confidential data and proprietary prompts. - Separate model capability from the surrounding product and engineering work. - Do not claim quality gains without explaining how they were measured. ### Clarifying Questions to Ask - What user or operational problem justified AI? - What evaluation set or review process checked output quality? - Which failures required guardrails or human judgment? ```hint Draw the boundary Make clear which decisions the model proposed and which decisions the system or a person controlled. ``` ### What a Strong Answer Covers - Problem selection and why an AI component was appropriate. - Your work on data, prompting or training, integration, evaluation, and safeguards. - Concrete failure modes, fallback behavior, privacy or cost considerations. - A measured result and a candid account of remaining limits. ### Follow-up Questions 1. What would make you remove the AI component? 2. How would you detect quality drift after launch?