Hardware, mostly — structures, avionics, a bit of control theory, and one piece of physics that turned into a paper.
The games live on their own sheet. This is the half that has to work: what the thing is, what I did on it, and what state it's in right now. Drawings are mine and deliberately schematic — enough to show the idea, not enough to build from.
Spin a wheel one way and the satellite turns the other. That's the whole trick, and it means you can point a spacecraft without spending propellant you are never getting back. I work on the wheel end of it — rim geometry, where the mass wants to sit, and keeping the thing balanced well enough that it doesn't shake the rest of the satellite apart once it's up at speed.
The idea behind fly-by-feel is a wing that knows what is happening to it — sensors living in the structure, reporting what the air is actually doing, instead of inferring all of it from a couple of air-data probes. My end is the control side: real-time state estimation and feedback, plus a lot of reading through what people have already tried so we're not quietly redoing it.
Role
Research assistant
Lab
Structures & Composites Lab
Team
5 PhD & postdoc researchers
Tools
MATLAB, embedded systems
Focus
State estimation, feedback control
Fig 03 · coupon in the rigFig 03 · elevation · NTS
PRJ-03
Composite fatigue testing
ongoing
Load a composite coupon, unload it, and do that a great many times over until it starts coming apart. I run the tests on Boeing coupons and watch where cracks start and how fast they travel, then pull the sensor data into MATLAB to map how strain and stress shift around over the life of the part. Less glamorous than it sounds, and genuinely interesting the moment a crack picks a direction.
Role
Research assistant
Lab
Structures & Composites Lab
Article
Boeing CFRP coupons
Loading
Cyclic, to crack growth
Analysis
MATLAB · strain/stress mapping
Fig 04 · airframe in assemblyFig 04 · plan view · NTS
PRJ-04
Tornado UAV airframe
iterating
The whole airframe built as a parametric Fusion 360 model, so that when somebody swaps a battery or moves a servo the rest of it follows instead of quietly stopping fitting together. Most of the work is a weight-against-stiffness argument with manufacturability as the tiebreaker — there's not much point in a layout nobody can actually build. Leaning on generative design took 28% off the mass without losing either.
Role
Core structures
Team
Stanford Flight Club
Tools
Fusion 360 · generative design
Mass
−28% vs. first layout
Trades
Weight, stability, manufacturability
Integrating
Avionics, transmission, controls
Fig 05 · on the padFig 05 · side elev · NTS
PRJ-05
High-power rocket · L2
in progress
A 3.9 in airframe on a 54 mm motor, built to certify Level 2. OpenRocket first, to check the stability margin and roughly where it should top out, then laser-cut and 3D-printed parts, epoxy, and a carbon layup on the fins. The avionics bay is mine as well — flight computer set up for dual deploy, so the drogue comes out at apogee and the main waits until it's much closer to the ground.
First-principles work on what happens where graphene meets aluminium, and whether doping that interface buys you anything worth having. Short version: of four candidate 2D reinforcements I simulated, two bond about 20× more strongly than graphene itself — the sort of result you hope for and don't really expect. It ended up in RSC: Materials Advances, which still feels a bit unreal.
Every drawing on this sheet is schematic and not to scale — they're here to show the idea, not to build from. Photos go in as I take them; until then the linework stands in.
If you want the real detail on any of these, email me and I'll happily talk your ear off.
