about
It started with a robot
The resume version: computer science student, software developer. The longer version starts on a competition robotics team, where I found out the code I wrote could send a very real machine across a field; and that I wanted to keep chasing that feeling.
The stack has changed since, but the thread hasn't: understand the system from first principles, then make it do something in the real world. Here's how that has gone so far.
the story so far
Four chapters, one thread
frc · team 1086
Make it move
I got into computer science through FRC Team 1086, where I was handed the part of the robot you can't see: the code. I wrote robot code and computer-vision pipelines, and built on-the-fly A* path planning so the robot could plot its own route across the field mid-match. Watching a graph search I wrote steer a real machine was the moment programming stopped being abstract for me.
ml · first principles
Understand the magic
When machine learning pulled me in, I didn't want to just call an API — I wanted to know what happens inside. So I built a GPT from scratch, working through the NeetCode machine-learning course submission by submission until every layer was something I had implemented myself. It's a slower way to learn. It's also why I trust what I build.
vision · ros 2
Teach it to see
The robotics thread and the ML thread merged in computer vision. I've built ROS 2 vision nodes and generated synthetic training data with NVIDIA Isaac Sim for pose estimation — because when real labeled data is scarce, you render your own. Perception is my favorite kind of problem: software that has to cope with the mess of the physical world.
web · full stack
Ship it for real
Software isn't finished until someone else can use it. I build for the web with TypeScript and Next.js, and in the CIT program I stood up a full-stack app the unglamorous way — Windows Server, IIS, SQL Server — and learned what deployment actually costs. That end-to-end habit follows me into every project.
values
How I work
Understand it from first principles
I'd rather build a GPT from scratch than call an API and move on. Knowing how the layers underneath work is what makes the layer on top reliable.
Software should meet the real world
The projects I care about most drive robots, process camera frames, and run on real servers — code whose output you can watch move.
Ship, then sharpen
A deployed, working version beats a perfect plan. I iterate in public on GitHub and improve from real feedback.
what's next
Where I'm headed
Open to software engineering internships
I'm early-career and deliberate about what that means: I'm looking for a team where code review is taken seriously, interns are trusted with real problems, and “works on my machine” isn't the finish line. I do my best work near the boundary where software meets the physical world — perception, ML systems, robotics — but I care more about shipping with people I can learn from than about any particular stack.
If your team builds things that have to survive contact with reality, I'd like to hear about it.
currently learning
On the workbench
What I'm actively studying right now. The list rotates; the habit doesn't.
- Transformer internals & training dynamics
- ROS 2 navigation stacks
- The math behind ML — linear algebra at J. Sargeant Reynolds
off the clock
Away from the keyboard
Away from the keyboard I'm playing soccer, out with friends, or hunting for music to focus to — almost everything on this site was built with headphones on.