Hi, I'm Cameron! I'm a rising junior at Harvey Mudd College studying CS and Math. I grew up in Northern Virginia, so now my favorite things include urban planning, summer thunderstorms, and good guacamole. (Only the really really good guacamole. I'm very picky about it.) My day-to-day interests include coding, talking about DnD, and excitedly thinking about how good sad music is.
Click an image to learn a little more about me :)
Get to know me better:
These are my cats back home, Siri and Floof. They are the most important thing about me. You might as well not read anything else on my about me pages.
Uh, anyway, they're the best. Siri is the sleek, stylish one and she's very smart. Floof is... the one who looks like he's named Floof. (He's silly and sweet.) Floof is the annoying younger sibling, and Siri is the annoyed older sibling.
Click an image to learn a little more about me :)
I've probably said this already, but I'm such a huge fan of good sad music and good character writing. I'm also a big musical fan and have a complicated relationship with Big Brother and Survivor.
Click an image to learn a little more about me :)
Some things I do when I'm not doing math or ranting about amazing music. I can never get enough of my friends or of traveling.
Or sometimes of whatever game I'm playing (the worst recent culprit has been Ticket to Ride). If you see the Ticket to Ride photos and you're sad that that bot is pretty bad... well, one day I decided I agreed
Click an image to learn a little more about me :)
HMC is a small STEM liberal arts college of about 900 students. I'm majoring in CS and Math, positioned happily between fun conceptual insights and rewarding real-world applications.
There are always cool problems to be solving at Mudd, and it has a famously challenging curriculum. I love a good challenge, though. I've maintained a 4.0 GPA even though only 14 people have ever graduated with a 4.0, and even though the freshman/sophomore core classes are usually the GPA-killer classes.
The best part of Mudd is being surrounded by people who are so genuinely excited by everything they do. Mudd is such a collaborative environment that everyone's passion for their work is really contagious. For instance, Real Analysis is quite a challenging class, but thinking through proof concepts with classmates is such a fun experience. I was scared to take it, but it ended up being one of my favorite classes I've ever taken. (Check out the pictures below for another reason why!)
Click a photo to learn a bit about me at Mudd :)
TJ is a highly selective STEM magnet school in Northern Virginia. The curriculum pushes you hard. Considering all the chaos that comes with high school, I'd honestly say it was even harder than Mudd. But it was also where I really fell in love with CS (math had already had my heart by then).
Like Mudd, getting all-A's is super hard, but I did. (Well, except senior year Geosystems where my A slipped to an A- in the fourth quarter, but that was intentional... because I wanted to prove to myself that it was okay to be imperfect....)
Click a photo to learn a bit about my high school experience :)
A genetic algorithm that evolves Ticket to Ride AI players. Trains a population of strategy-weight vectors through selection, crossover, and mutation across multiple generations; evolved bots outperform the game's built-in opponents.
A backtesting engine from scratch using only NumPy, including transaction cost modeling, momentum / mean-reversion / cointegration-based pairs strategies, and from-scratch ADF / Engle-Granger testing. Uses walk-forward assessments and Sharpe ratio parameter sweep heatmaps to assess out-of-sample overfitting.
Implements Bayesian optimization from scratch using only NumPy. Fits a Gaussian process posterior via an RBF kernel and Cholesky decomposition, then selects the next sample point via Upper Confidence Bound (UCB) or Expected Improvement acquisition functions.
Prices European options under both Black-Scholes and Monte Carlo simulation, displaying results side by side. Adjust strike, expiry, and volatility to see how the simulated price distribution and theoretical value respond in real time.
Extended an existing production codebase at Fermat into an interactive dashboard for remodeling industry loss warranties (ILWs), constructing exceedance probability (EP) curves from user-defined combinations of catastrophe scenarios with custom trigger logic, querying MSSQL databases live in place of manual querying for risk analytics team.
Built a web application to host custom AI agent workflows for the Fermat R&D team. Integrated the Anthropic API via AWS Bedrock with Gmail, Drive, and Box APIs. Balanced LLM inference with lower-cost algorithmic approaches and designed custom processing pipelines for tasks exceeding the model's capabilities.
Implemented feedforward neural networks and backpropagation from scratch three times across coursework: twice in pure Python with no external libraries (including manual matrix operations), and once with NumPy, trained on MNIST handwritten digit classification. Supported configurable network depth, layer widths, and activation functions.
Applied skewed probability distributions to the feature-splitting process within individual decision trees, introducing controlled randomness to improve Random Forest ensemble stability.
Designed and trained a custom LSTM model to predict priority and category labels for M.C. Dean database entries with highly variable formatting and content inconsistencies.