This section captures how I think about problems — and what I think about.
The three parts connect: start with the operating principles, see where they lead in terms of research interests, and then follow the open questions that haven’t resolved yet.
The mental models and instincts I keep returning to — physics-first decomposition, failure-mode analysis, timescale separation, and why AI is most useful when grounded in structure.
Start here if you want to understand the reasoning behind the projects.
The technical areas I find genuinely interesting — not just what I’ve worked on, but what I keep thinking about.
Covers: dynamics and instability, contact mechanics, Hamiltonian formulations, evolutionary algorithms, physics-informed ML, and the patterns that keep appearing across domains that look unrelated on the surface.
Continue here if you want to see where the thinking leads technically.
Ongoing directions that haven’t resolved into finished projects yet — half-formed hypotheses, active explorations, and the questions that keep coming back.
The unresolved end of the thread.
Physics First Most problems I work on start with the governing equations, not the data. Before reaching for a model or a tool, I try to understand: what forces are acting, what constraints exist, where the system can fail, and how it behaves at its boundaries.
This isn’t dogma — it’s practical. A model built from physical understanding generalises better, degrades more predictably, and can tell you why it’s wrong when it fails. A model built from curve-fitting can’t.
...
These are the things that haven’t resolved yet.
Not finished projects — those live in Work. Not settled positions — those are in Research Interests. This is the frontier: directions being actively explored, hypotheses not yet tested, questions that keep surfacing from different angles.
🤖 AI-Assisted Simulation Debugging Current simulation workflows fail silently or produce results that look plausible but are physically wrong. The debugging process is manual, slow, and heavily dependent on domain expertise that isn’t transferable.
...
Most of what I find genuinely interesting in engineering sits at the intersection of two things: the governing equations that describe a system, and the behavior that emerges when you push that system toward its limits.
The stable, well-behaved middle of the operating envelope is well understood and usually boring. The interesting physics — the instabilities, the mode transitions, the nonlinear couplings — lives at the boundaries. That’s been a consistent thread across tires, piston rings, batteries, and AI systems. Different physics, same instinct.
...