teaching
Teaching philosophy, mentoring, and courses
Teaching and research as one activity.
I view teaching and research as deeply intertwined. The same autonomous workflows I build for materials discovery are, fundamentally, structured ways of learning — and the classroom is where I help students develop both the technical foundations and the scientific intuition needed to drive that kind of work.
Training students across mechanics, materials, computation, and AI.
Many of my mentees have co-authored peer-reviewed publications spanning DFT, machine-learned interatomic potentials, phase-field modeling, and agentic AI for scientific simulation. Mentoring — from first-year graduate students to accomplished high-school scientists — is one of the parts of academic work I find most rewarding.
K–12 outreach — NAACP ACT-SO Youth Program
- Kudzi Makoni (Neuqua Valley High School) — Silver Medal, Chemistry, 2026 NAACP ACT-SO National Competition; Bronze Medal, 2025.
Instructor and teaching assistant appointments.
| Role | Course | Term | Format | Enrollment |
|---|---|---|---|---|
| Co-Instructor | Kinetic Phenomena in Materials (elective) | Spring 2017 | In-person | 30 |
| Teaching Assistant | Solid Mechanics of Materials (core) | Fall 2016 | In-person | 72 |
| Teaching Assistant | Advanced Mechanics of Materials (core) | Fall 2015 | In-person | 76 |
Undergraduate core and graduate electives.
Undergraduate core
- Materials Science and Engineering
- Solid Mechanics / Mechanics of Materials
- Thermodynamics of Materials
- Numerical Methods for Engineers
Graduate / advanced electives
- Computational Materials Science
- Density Functional Theory: Theory and Practice
- Machine Learning for Materials Science
- Multiscale Modeling: From Quantum Mechanics to Continuum
Using AI to teach — thoughtfully, and with the science front and center.
The same technologies I research are reshaping how students learn engineering and materials science. I incorporate them into my teaching not as substitutes for fundamentals, but as tools that let students engage with harder problems, sooner.
AI as a study partner, not an answer key
Students use large language models to explain concepts back to themselves, generate practice problems, debug derivations, and query the syllabus — but assignments require them to show reasoning, cite sources, and reproduce derivations by hand. I design problem sets that reward genuine understanding over polished output.
Computational assignments with AI-assisted coding
In computational courses (DFT, MD, phase-field, ML for materials), students use AI coding assistants as pair programmers — accelerating boilerplate so class time focuses on numerical methods, convergence testing, physical interpretation, and error analysis. Rubrics evaluate scientific correctness and understanding of the underlying methods, not lines of code produced.
Live demonstrations of autonomous scientific workflows
My research platforms — including AutoMOOSE and TEM-Scientist — become live classroom demonstrations of how AI agents interact with real simulation and experimental tools. Students see both the capabilities and the failure modes, and learn to reason about when to trust an autonomous system.
Explicit conversations about AI literacy and integrity
Every course I teach opens with a clear discussion of appropriate AI use: what counts as collaboration, what counts as academic dishonesty, and how to evaluate AI-generated content critically. Students graduate knowing not only how to use these tools, but when not to trust them.
A graduate course only this program could offer.
AI and Agentic Systems for Scientific Discovery
A graduate course on how modern AI — large language models, autonomous agents, reinforcement learning, and physics-informed neural models — can be composed into trustworthy scientific workflows. Students would work with real simulation and instrument-control codes, learn to design agentic architectures that retain deterministic execution and verification, and confront the epistemic questions autonomous science raises (what counts as evidence, what an agent should refuse to do, how uncertainty should propagate).
The course draws directly on the platforms my own research develops — including AutoMOOSE and TEM-Scientist — and prepares students to build the next generation of autonomous scientific tools across mechanics, materials, chemistry, and engineering more broadly.