Background

I’m in my final semester of a 5th year masters in Mechanical Engineering at Carnegie Mellon University in Pittsburgh, PA. I primarily focus in robot systems design and motion control systems (GNC), including optimal control theory and deep RL based control.

Skills

Software			Hardware
Frontend	React		Computer Aided Design (CAD)	Onshape
Backend	Python Julia Lang		PCB Design	KiCad
Embedded	C++		Design For Manufacturing + Assembly	3D Printing CNC Machining Lathe Water Jet Sheet Metal Bending
Optimal Control / RL	Julia Lang Python
Deep RL	Python
Algorithms (LP / QP)	Interior point methods LQR CVX MPC
Algorithms (NLP)	iLQR DIRCOL
Algorithms (DRL)	PPO TD3

Education

M.S. Mechanical Engineering - Carnegie Mellon (Expected graduation May 2025)

B.S. Mechanical Engineering - Carnegie Mellon

Experiences

Dirac Inc. - Backend Engineer (2024)

Dirac is a startup in NYC/LA that is creating a product to automate work instructions for manufacturing purposes. During my internship here, I developed a trajectory optimization algorithm that would create minimal collision paths for general 3D objects to travel during the assembly of a system. I also developed an custom signed distance field (SDF) solver that would generate data structs for a given 3D object at a desired resolution.

Language: Python

Ford Motors - Computer Vision Research Engineer (2023)

During the manufacturing process of a vehicle, it's critical to inspect the installation of components at various points along the line in order to catch errors early and minimize the lost costs of correcting assembly errors. Ford is exploring the use of deep neural nets to capture images of their vehicles, detect sub assemblies, and classify if the installation has been successful. A huge bottle neck in this is the need to capture and label hundreds of images in order to train a model to accomplish this detection/classification task. During my internship at Ford, I developed a python tool which enables users to load in a CAD model of the target assembly, and generate a fully labeled synthetic image data set for training a model to complete this task, eliminating the need for a human to complete the arduous capture / label process. I found that through sufficient domain randomization, it was possible to train a model on 100% synnthetic images to achieve >90% accuracy when deployed in the real world. Critically, this now allows engineering teams at ford to have a model pre-trained, ready to deploy as soon as a line change occurs, as opposed to deploying the assembly line change followed by image collections, training, and testing tasks.

Language: Python

CMU Biorobotics Lab - Research Assistant (2021-2022)

During my time with the Biorobotics lab, I worked on a research team developing a robot to achieve pipe-in-pipe trenchless repair for the decaying natural gas piping infrastructure around the US. I designed and tested a radial nozzel that would extrude an exotic resin in a pipe shape creating a smaller pipe within the old infrastructure. This nozzle was design to adapt its exterior geometry to acomodate stress induced ellipsoidal fallicies in the host pipe geometry.

CU Boulder Matter Assembly Lab - Research Assistant (2020)

During my time with the Matter Assembly Lab, I developed a novel print-in-place pneumatic flex sensor for a soft robotic finger. This allowed for pose estimation of the actuator without the introduction of rigid sensing components in the actuator itself, preserving the desireable features of a soft actuated finger

United States Air Force - Communications (2014-2020)

In the Air force I worked as a communications technician. My responsiibilities included operating and maintaining SATCOM systems, as well as deploying tactical HF communication systems in various regions and terrains around the world, from mountains to forrest / jungle terrain. I completed a 6 month deployment to southern Turkey in support of mission Enduring Freedom. I honorably seperated from the Air Force at the completion of my 6-year enlistment as a Staff Sergeant (E-5).