What this interview will probe
Build AI-powered features into Mercury's product, integrating large language models into workflows like transaction categorization, support automation, and financial insights while keeping outputs safe and reliable in a regulated banking context. You'd design prompts, evaluation pipelines, and the surrounding application plumbing in Haskell and TypeScript. A technical interview would probe how you structure LLM workflows, guard against hallucination and prompt injection in a fintech setting, and build evaluations to measure whether an AI feature is actually correct and trustworthy.
ExoForm is not affiliated with Mercury. This is an independent practice page.