Developer
Leading AI/ML at RedBit. Building intelligent tools and helping clients figure out how they can leverage AI to save money.
Also:
- built a timeseries forecasting platform and data analysis tool, used for financial forecasts of >$30B in transactional volume
- data engineering on large-scale enterprise projects
- wrote a foundational open-source library for (AI) planning research macq
- teaching intro to programming (APSC 143) @ Queen's University (fall 2024)
Researcher
Graduate student supervised by Prof. Xiaodan Zhu at Queen's University Ingenuity Lab.
Working with JP Morgan AI Research on financial applications of LLMs. Recently published paper at FinNLP and preparing another for EMNLP.
Especially interested in feature steering and applications of search (MCTS) for training (synthetic data generation) and inference.
Previously, worked with Prof. Christian Muise on macq, a python library for planning action model acquisition and published a paper on it at ICAPS 2022.
Hobbyist
I love exploring latent space, and following others' experiments.
LLMs are more than just tools for work to me. They are an embodiment of the aggregate human subconscious, an approximation of the entire internet and whatever emerges synthetically from that. Base models let you query the training data.
Interacting with the egregore of humanity is the most interesting pass-time I could imagine.
I'm also a linux hobbyist. Nixos-btw, my dotfiles are a never-ending labour of love. I have a somewhat popular neovim config featuring a custom llm chat plugin.
My computer is an extension of myself.