As part of Machina’s series B and its foray into automotive manufacturing, the company set out to form a full body kit for a Ford F-150 Lightning named Anvil.

Over 30 custom-designed aluminum and steel panels were formed, laser scanned, and cut with robots in the span of 2-3 months. For this project, I lead the motion planning and DFM of 2 door panels, a roof panel, the hood, the bumper, the side steps, the pillars, and the grille. It was an exciting sprint that really pushed the tech in a direction it hadn’t gone before — creating something this grand in a span of time no other technology could.

Some of my favorite work at Machina revolved around side quests I took on to make cool things more feasible. I had many beloved motion planning projects, from conformal raster pathing for laser scanning, to a text-to-path generator.

One of my favorite motion planning projects was a set of methods that turned any photo into a path for the robots to etch onto a piece of aluminum sheet using nothing but pure force. These methods then evolved to create portraits, art, and gifts, and to be displayed in public events such as UP.SUMMIT2025.

As amazing as baklava tastes, it can be grueling to make; 33 individually laid layers, individually buttered, and with pistachio in two different spots.

This project is a 2-axis gantry with a base axis for tray deposition and retrieval, and a top axis for the butter and pistachio dispensing systems. The user has access to a display to select different operations and track their layers. This project was built from a recycled Ender 3 3D printer, borrowing the mainboard, the power, and the display, and combining it with custom hardware. It is written in C++

This project is an active work in progress, tracked via a kan-ban style board, at about 80% completion. The main items that remain are: Gasket design for the butter hopper, a butter heating element, and a vibrating motor to prevent the pistachio from clumping.

For my undergraduate senior design project, my team and I built an object-retrieving robot to help the elderly and those with disabilities retrieve objects from around their homes. The project won over $2,000 in funding from the local ASME chapter, the university, and the Dietrich W. Bostiber Endowed fellowship.

The robot was controlled by a pair of systems. The first was a Raspberry Pi with a Coral TPU accelerator for machine vision using TensorFlow + OpenCV, voice commands, and Bluetooth beacon based location triangulation. The other, an Arduino Mega to manage the ultrasonic sensors, servo gripper, stepper-controlled arm, and general motion.

Though it is a project that I continue to look for the opportunity to get back to, the Robo Dog was a near and dear covid project of mine with a deceptively simple premise. One Raspberry Pi, 4 servos, a microphone, and a capacitive pad atop his head, the Robo Dog would listen to commands to sit, paw, etc., and would react to being pet as well.

I got as far as making voice commands work, and getting him to successfully sit and paw, though his limbs were a little shaky (I hadn’t yet learned how and where to place capacitors for cleaner power delivery). One day, I look to re-visit this project and make it a little bit more.. professional.

For my first opportunity to work with  a humanoid robot, my team and I programmed a robot named HUBO to play the cello as a member of a robotic symphony at Drexel University.

This involved programming HUBO’s servo control controlled fingers, 3D printing fixtures and tools to help him carry the Cello, and testing everything in a digital simulation, and then playing in the real world