About the course
This graduate course treats fluency with generative AI as a research skill. Students learn to work with AI agents—systems that read your files, run code, search the literature, and carry out multi-step tasks—and to bring them into their own research: reviewing and synthesizing literature, writing and reasoning about code, analyzing data, and organizing what they know. This is not a computer science course, and the goal is not to build AI systems; it is to use them well, with enough understanding of how they work to use them critically.
The course is open to graduate students, at the PhD or master's level, across engineering and the sciences. If your research involves reading a literature, writing code, or working with data, it is meant for you: students from computational biology, physics, data science, and other programs are welcome alongside engineers. Your own research is the material; assignments are applied to the problems you are working on, not to toy examples.
What you will learn
- Working with AI agents
- From the first weeks, you will use agentic tools in chat interfaces, coding environments, and notebooks, and learn what to delegate, how to direct the work, and how to check the result.
- Context engineering
- The skill underneath every tool: deliberately designing what an AI system sees, remembers, and is asked to do. The same discipline shapes a good prompt, a project instruction file, a reusable agent skill, and a well-organized research project.
- Literature and research knowledge
- AI-assisted discovery and synthesis of the literature, deep-research agents, and ways to build a personal knowledge base that compounds over time instead of being re-derived with every query.
- Building your own research tools
- Small automations and agent skills tailored to your field, built with AI assistance even if you are not an experienced software developer.
- Critical judgment and responsible use
- How language models work, at the depth a researcher needs; the illusions of understanding that AI can foster in science; and the practical questions of research integrity, including the disclosure policies of journals and funders and what advisors and committees can reasonably expect.
How the course works
The class meets weekly for a two-and-a-half-hour session that combines discussion with hands-on work, and extensive lecture notes accompany each session. Over the semester, each student builds a working setup for their own research: instruction files, agent skills, and organized notes that stay useful after the course ends. No prior experience with git or GitHub is needed; version control is introduced in class as one option for managing that work.
What is new in 2027
MAE-6291 first ran in the spring of 2026 as a seminar for engineering doctoral students. The second edition is renamed and opened to graduate students across the sciences and engineering. It also follows the field: AI tools have moved decisively toward agents, even in the everyday chat interfaces of the major vendors, so the course now puts agentic work at the center from the early weeks rather than in the second half of the semester. A thread on responsible use, research integrity, and disclosure runs through the whole term.
Details
- Instructor
- Prof. Lorena A. Barba
- Course
- MAE-6291 (Special Topic), Spring 2027
- Meets
- Tuesdays, 12:45–3:15 pm*; room to be announced
*Meeting time subject to change. - Term
- Classes run January 11 to April 26, 2027; the first meeting is Tuesday, January 12.
- Who can enroll
- Graduate students (PhD or master's) in engineering or the sciences. Students outside the School of Engineering and Applied Science: please email the instructor before registering.
- Prerequisites
- Comfort writing and running Python scripts. No prior AI or machine-learning coursework, and no experience with git or GitHub, is required.
- AI tools and cost
- You can complete the course without additional cost for AI products. GitHub Copilot is free for students through the GitHub Student Developer Pack, and API access will be provided for class activities. Some students may choose a personal subscription to an AI assistant (typically about $20 per month) to work with a particular model; this is optional.
- Contact
- labarba@gwu.edu
Preview: the first edition
The Spring 2026 edition is openly archived under CC BY 4.0: lecture notes and slides for all thirteen sessions, a recording of the final class, and its slides on Figshare. The second edition will be revised, but the archived course page shows the approach and the depth of the material.