About

Hi, I'm Quilee.

I'm an Applied Scientist at Amazon FAI (formerly Amazon AGI). I work on the quality of the expert-sourced data and evaluations that frontier language models are trained and tested on. On most days I build agent-based pipelines and LLM judges that audit STEM expert data, calibrate those judges against human experts, and develop reinforcement learning environments for coding, STEM reasoning, and tool use.

I grew up in St. Lucia and came to MIT for a B.S. in Computation and Cognition. I stayed for graduate work in Ed Boyden's Synthetic Neurobiology Group and earned an M.S. in Computational Neuroscience. In the lab I built neural-network models of the C. elegans nervous system, measured how they scale with data and parameters, and assembled an open dataset that combines neural recordings from 11 public datasets (about 900 worms) with the available connectomes. I also spent a lot of time rebuilding a light-sheet microscope.

In 2025 I moved toward AI engineering. At Numenta I designed a way to grow a pretrained language model from 70 million to 1.4 billion parameters, starting the larger model from the smaller one's trained weights. At MIT's Office of Research Computing and Data, I supported researchers on the university's clusters and built tooling for running LLM agents on them.

On the side I build with coding agents. With Justin Wei and Yile Fan I built Octopus, a system in which a coding agent writes and tests the driver for whatever hardware is plugged in. At a Sundai hackathon, my team built GrowLab, a harness that lets a research agent search training schedules. I write about these projects on the blog.

Outside of work I enjoy reading fiction, hiking, and traveling to understand different cultures.

Community

  • Builder, Sundai Club weekly AI hackathons (2025 – Present)
  • Co-President, IEEE-HKN Beta Theta Chapter at MIT (2020 – 2022)
  • Secretary, Beta Chapter of Theta Chi at MIT (2017 – 2021)
  • Social Chair, MIT Black Student Union (2018 – 2019)
  • Mentor and advocate for open science and for underrepresented students in STEM