William H. English

Hello visitor! My name is Will English, and I’m a Computer Engineering PhD student at the University of Florida, where I work with Dr. Rickard Ewetz on neurosymbolic AI. I received my B.S. in Computer Science from the University of Central Florida in 2023.

My research lives at the intersection of formal methods and machine learning. I am interested in systems that unify rigorous symbolic reasoning with the flexibility of neural models. I’ve worked on methods for autoformalization of natural language into temporal logic, enforcing safety and correctness guardrails on language models and robots through novel constrained decoding algorithms, and recently I’ve had the opportunity to reverse-engineer how vision-language models implement spatial and temporal reasoning circuits thanks to my colleaue Chase Walker. In a nutshell, my work is about making neural systems more reliable, verifiable, and interpretable.

Additionally, I am (almost religiously) devoted to algorithmic optimization. A blog post is coming, but the takeaway is essentially this:

\[\text{Cost} = \underbrace{\frac{\$}{J}}_{\text{electricity}} \times \underbrace{\frac{J}{\text{FLOP}}}_{\text{hardware}} \times \underbrace{f(n)}_{\text{algorithm}}\]

Energy, time, and money are not three separate costs. They’re one cost expressible in three unit conversions, all linear in the number of operations $f(n)$. So when you reduce an algorithm’s complexity, say from $O(n^2)$ to $O(n \log n)$, you don’t just save time: you save energy and money by the exact same factor. The other two terms are fixed by your electricity provider and your silicon; the algorithm is the one you control. So it must be optimized, in the name of efficiency. If you could fill up your gas tank cheaper by making the fuel pump faster, you’d be a fool not to make that choice! Originally from Pittsburgh, PA (go Steelers), I grew up in Saint Augustine, Florida. Outside of academia, some of my interests are backpacking, algorithmic composition (music from code), books (reading and listening), game-making, lucid dreaming, occultism, and all kinds of fun sidequests.

This site serves as a CV, portfolio, blog, client portal, and repository of information and tools I find useful. If you’re looking for a tutor, please visit my teaching page. Otherwise, feel free to explore and learn about my professional and recreational pursuits.