Mechanical engineer · Controls researcher · Prototype designer

Make it.
Learn from it.
Repeat.

I move quickly between physical mockups, visual stories, sensing, and code—turning early questions into prototypes that can be tested, used, and refined.

See the interaction studies View CV
HoursQuestions become something you can touch 10 weeksMechanical, sensing, and software become one robot < 2 weeksA hand sketch becomes a functional native iPad app OngoingA software frustration becomes a native product
Cardboard mockup + animated interface. Touch cannot be inferred from code alone.

A prototype is not a miniature final product. It is a question with a physical form.

Interaction study · Aug 2026

Disc: a haptic display concept

A palm-sized companion for surfaces: a display you can rotate, press, and feel without turning every interaction into another phone session.

Hand-drawn Disc concept exploring placement, controls, timers, music, Siri, and calendar interactions
The first sketch in my design notebook revealed placement, behavior, and use cases together. Reviewing it prompted the next question: How can I get a better feel for this? Within a few hours, I built a cardboard sketch model. It answered some questions and raised others, so I animated a short input-action loop in Keynote.

Two more open questions

Sketch broadly before committing deeply.

These are deliberately early. Each sketch identifies a behavior worth testing rather than pretending the implementation is already resolved.

Hand-drawn Orb concept using translucent color, touch, rotation, and motion for ambient notifications

Orb · Ambient information

Can a device communicate without demanding a screen?

A softly matte, translucent surface uses color, pulse, touch, and rotation to express music, timers, calls, and home alerts.

Next prototype: Compare light alone with a rotating or vertically actuated form. Which motion reads as useful rather than distracting?

Hover Touch sketch using a front camera and projected shadow to estimate a gloved fingertip above a phone

Hover Touch · Robust input

What happens when the screen cannot trust a touch?

For gloves or wet conditions, a camera-estimated fingertip casts a responsive shadow; moving toward the display becomes a tap and moving laterally becomes a drag.

Next prototype: Separate hand jitter from phone motion, then learn which visual, haptic, and audio cues make an “air tap” feel dependable.

Integrated prototype · 10 weeks · Team lead

C.L.E.A.N.: from weekly prototypes to one autonomous robot

Five people combined vision, motion, sensing, and grasping into a robot that could find, approach, verify, and lift an aluminum can.

C.L.E.A.N. project poster showing the robot, computer vision, mechanical system, electronics, and results
5person team
~1,500training images
Weeklybuild–test cycles
  • Collaborated with a five-person team through weekly design-build-test cycles, working across the mechanical, electrical, and software interfaces.
  • Trained and tuned an OpenCV cascade classifier, then converted a detected can’s image position into steering commands on a Raspberry Pi.
  • Integrated custom printed treads, motor drivers, ultrasonic and inductive sensing, and a 2-DOF worm-gear arm and claw.
What this project shows

I can keep a longer prototype moving: define the next useful test, help specialists connect their work, and carry an idea through to a functioning physical system.

Rapid prototype · < 2 weeks from sketch to app

Frame: a spare iPad becomes a calm window into the day

A design-first dashboard for the room—and a study in moving rapidly from hand sketch to interaction mockup to working native software.

Frame dashboard showing a real Doorbell camera feed alongside HomeKit controls

Frame treats a persistent iPad as household infrastructure rather than another destination to open.

  • Moved from hand sketch to Keynote/Figma mockup to a functional native app in under two weeks.
  • SwiftUI dashboard built around a large, swipeable feed carousel.
  • WeatherKit, EventKit, HomeKit, AVFoundation, HealthKit, and MusicKit integration paths.
  • Explicit loading, denied, stale, offline, empty, and unavailable states.
Open Frame case study

Native software product · Ongoing

Raagtex: software is a prototyping material too

A native Apple-platform LaTeX cockpit built around one product question: can technical writing, compile feedback, and PDF preview share one calm workspace?

Raagtex on macOS with project navigation, LaTeX editor, controls, and live PDF preview
A functional SwiftUI app for macOS and iPad—not a static interface rendering.

I started Raagtex because many LaTeX tools make the writing-to-PDF loop feel more complicated than the work itself.

  • Native SwiftUI interface across macOS and iPad.
  • Local project browsing, editing, compilation, diagnostics, and PDF preview.
  • Multiple layouts and visual modes explored as interaction choices, not decoration.
View Raagtex View source

For the technical reader

Systems & autonomy

These projects use control theory as behavior design: estimate what a system is doing, choose an action, and recover when the world does not behave exactly as expected.

Swing-up, balance, and catching behaviors for an inverted pendulum robot

Physical robotics

Inverted pendulum: recover, balance, and return safely

Identified a physical system from experiments, estimated its state, and built a mode-switching controller that could swing up, catch, balance, and return to rest.

Technical project →
Simulated satellite rendezvous trajectories for LQR and model predictive controllers

Constrained autonomy

Satellite rendezvous: plan motion inside real limits

Compared autonomous guidance strategies across uncertain starting conditions while balancing convergence, fuel use, time, and actuator constraints.

Technical project →