BS in Mechanical Engineering, UCSB, expected 2023
Hi, my name is Flynn and I am currently studying mechanical engineering at UC Santa Barbara. I have worked with a machine learning group on campus, completed an internship with Chevron, and maintained a 4.0 GPA as a Regent's Scholar and Honors Engineering Student. I am fascinated by audio and design.
Machine Learning Research at UCSB
JANUARY 2020 - JUNE 2020
Simplification and pruning of a system identification neural network
This past school year, I worked with the Moehlis Research Group at UCSB to create software in Python to automatically simplify and prune the output from the system identification neural network they were developing. This machine learning algorithm provided an exciting new way for scientists to create models for their datasets by providing the accuracy and adaptability of machine learning while avoiding it's traditional pitfall of obscurity. While most existing techniques can create a model that provides accurate predictions, they often speak little to the physics or biology behind the data. On the other hand, the UCSB algorithm provides a mathematical formula based in real principles. This algorithm employs a "grey box" approach while most existing techniques behave as "black boxes."
My role in the project was to automate the conversion of the weights of the neural network into readable formulae. This process involved two steps, namely simplification and pruning. The first step, simplification, involved converting the weights into equations. This was fairly straightforward. The challenging step, pruning, involved attempting to combine these hundreds or even thousands of equations into one readable equation that accurately fit the provided dataset. This involved rounding numbers so exponential, sinusoidal, and sigmoidal functions would combine while avoiding loss in accuracy so the final equation would still fit the data. Coming into the project with no coding experience in Python, I definitely learned a lot both about automation of human intuition and machine learning!
Internship at Chevron
JUNE 2020 - AUGUST 2020
Self-taught new skills in control systems, instrument selection, and economic analysis
Co-led an eight-member intern team, despite being the youngest intern
Praised by supervisor for a “creative and unique [subsea architecture layout] that reduced downtime, expensive directional drilling, and complexity”
Despite the circumstances and sudden switch to a virtual internship, I had a great time this summer working with a team of eight interns. I was able to branch out beyond my mechanical engineering training and work on control systems, instrument selection, subsea architecture design, and economic calculations like Net Present Value for the first time. I enjoyed working with a small team on big projects and learned to get scrappy with my knowledge, applying what I knew from school, my own hobbies, my mentors, and the internet.
"Flynn took the spotlight in the last hour of this project when the team’s mentors threw a curveball, forcing the team to quickly redesign the steam flash drum’s level alarms and controls. He efficiently and effectively led his team to a new design while developing a new proportional derivative control scheme."
-Supervisor, Lead Process Engineer at Chevron
MIDI Drum Prototype
MAY 2020 - JUNE 2020
A light up MIDI drum that can trigger kicks, snares, vocal chops, or anything at all
This year, in ME10, we were tasked to model a novel product in SOLIDWORKS, create a prototype, and code it using Arduino. I decided to build something I had been thinking about for a while, which was a light up drum that could trigger sounds on a computer. Follow the link below to see a full description of the project!
2014 - PRESENT
I have been writing and producing music since the 8th grade. I love both the musical side of composition and the technical aspects of post-production.
Recently, I started a new project with Alex Cronin, titled Havey. Our debut EP, aptly named "EP001," will be released in January. Our sound focuses on powerful melodies and intricate drops, rooted in electronic and pop music with synthwave, rock, and orchestral influences.
You can listen to one of the singles below:
JUNE 2019 - AUGUST 2019
Last summer, I had the wonderful opportunity to work with Leo, a middle school student hoping to attend NASA's Space Camp. Using the Roomba Open Interface paired with an Arduino, I helped him modify his iRobot cleaning robot to detect high concentrations of 'methane' on the surface of 'mars.' I got to teach him skills including Arduino syntax, debugging, and dead reckoning. A few months after working with him, we got the amazing news that he received a full ride scholarship at the camp!