Mastering AI & Building Tools for Research and Professional Success

4 Credits.

Large language models are the most consequential professional tool to emerge in generations. LLMs are rewiring the daily practice of every profession that runs on reading, writing, analyzing, designing, and building through code. Students who master these instruments, and who can spot the exact points at which they fail, will hold a durable advantage in any field they enter. This seminar is a practical guide to that mastery, grounded in both theory and practice. The course is taught in two halves.Part 1 — Foundations (Thomas Rid). The first half builds the conceptual scaffolding: what language models are, mechanically and mathematically; how frontier models are trained, scaled, aligned, and evaluated. Sessions cover model capabilities and failure modes, prompting as a craft, verification and source-checking workflows, interpretability research, and the hard problems along the way. Because the frontier moves faster than any syllabus, this half also functions as a live laboratory in which students test cutting-edge tools as they ship, including releases that postdate the first day of class. The goal is calibrated trust: knowing when to lean on a model, when to second-guess it, and when to put it down.Part 2 — Applications (Lee Foster). The second half moves from understanding models to deploying them. Students will work with agents and multi-step pipelines: LLMs operating as components inside larger systems. Sessions are hands-on and oriented around the tasks professionals actually face: deep research, document analysis, coding, quantitative work, structured brainstorming, reading at scale, and writing under deadline. The half culminates in a semester-long project in which each student builds a working AI tool or pipeline addressing a problem from their own domain. The project is the principal assessment.