TOUGH TONGUE AI
  • Home
    LogoTOUGH TONGUE AI

    RESOURCES

    Loading...

    HomeLibraryFDE Mock · LLMs and RAG

    FDE Mock · LLMs and RAG

    FDE Mock · LLMs and RAG

    Overview

    The AI deployment round with Ryan. A bank wants an assistant over its documents: how do you build it, what goes wrong in the pilot, and what do you tell the customer? Written report after, no hints.

    Your Instructions

    Your goal: show that you can take a customer from "we want an assistant over our documents" to a pilot that works, and say clearly what you chose and why.

    How this works (about 30 minutes)

    • Ryan asks one main question with a made-up customer, and one shorter question. He answers questions about the customer; he will not volunteer facts.
    • Think out loud. Ask for what you need.
    • A notepad is available. Ask Ryan to open it if you want to sketch.
    • Lots of follow-ups and one curveball. No hints or scores during the call.

    What scores well

    • Starting from the users and the questions they ask, not the tools
    • A clear picture of the pipeline: documents, chunks, retrieval, the answer, the sources
    • Knowing where wrong answers come from and how you would find the failing step
    • Saying how you would test it with the customer before the pilot grows
    • Calm, plain language a bank manager could follow

    Tips

    • Take FDE Coach · LLMs and RAG first.
    • Say what you would build first and what you would leave out.

    More in FDE Interview Prep: AI Coach Edition

    • FDE Mock · Build a New Project
    • FDE Coach · Extend and Debug
    • FDE Mock · Extend and Debug
    • FDE Coach · AI-Assisted Coding
    • FDE Mock · AI-Assisted Coding
    • FDE Coach · LLMs and RAG
    • FDE Coach · Evals and Safety
    • FDE Mock · Evals and Safety
    • FDE Coach · The Customer Room
    • FDE Mock · Customer Room: Scope Push
    • FDE Mock · Customer Room: Explain It to the COO
    • FDE Coach · Project Deep Dive
    View the full course (43 sessions) →