ComfyUI / VFX

I worked a lot with ComfyUI this summer on the first version version of Bronte, which was supposed to be a Runway Aleph-style tool. I didn’t really have much to show for it, because I wasn’t happy with the results, but I wanted to share my thoughts here because I definitely learned a lot.

Poker Coach

Little webapp I made to learn GTO-ish poker. Lots I want to add / fix, but I thought I’d share it here. The repo is public if anyone wants to contribute!

Bronte

My main focus right now is Bronte, an AI agent for the VFX industry. Here’s our website, going to start beta testing the Houdini plugin over the next few months!

Weaver

Check out my LLM-based storybuilder website! Would love feedback, have some more features I want to implement soon.

Rendering Engine

I’ve been building a C++ / OpenGL-based rendering engine to learn GPU and graphics programming. This repo contains some OpenGL code and thoughts on graphics in general.

Graph neural diffusion

This repo contains the work I did for my honours project, where I studied the GRAND GNN architecture, specifically evaluating the claim that their model is especially resistant to oversmoothing.

Starting to work on the Vesuvius Challenge

I’ve been developing an autosegmentation tool as part of the Vesuvius Challenge. It uses DBSCAN for initial segmenting, and I’m currently working on implementing a step to correct low-confidence regions according to data from high-confidence regions using Bayesian inference.

An n=1 trial on the effects of Vyvanse

I ran a 30-day trial to figure out the effect my ADHD medication has on my productivity and mood.

Lead-acid battery recycling and Dodd-Frank 1502

This is a piece I wrote about the relationship between demand-side regulation like Section 1502 of the Dodd-Frank act and lead policy.

Predicting Population Size

This is a program I wrote in Python to answer questions of the type “If I sample from a pool 100 times, and 1 object appeared 3 times, 2 objects appeared twice, and every other object appear once, what is the most likely size of the pool?” (which I later learned biologists know as the mark and recapture process to estimate population sizes). It uses a Monte Carlo algorithm and recursive search to estimate the most likely pool size your sample came from.

The Angel Problem

I got an undergraduate research grant to study the angel problem under Nadia Lafrenière and Assaf Shani. I’ve developed an interactive tool in Python to explore the problem and illustrate parts of the proof.