I aided the DeepMind music team in their research on generative models for music. We prototyped integrations between their music models and my music studio and Digital Audio Workstations like Ableton. We've collaborated with musicians across genres like Jacob Collier. You can explore our work here: deepmind.google/models/lyria/realtime

I wrote a policy paper on Refocusing Higher Education on Career Outcomes based on my work with Make School. The research was published by the Manhattan Institute as part of it's "Beyond the Beltway" series looking at innovative ideas in higher education from outside of DC. The paper lives on the Department of Education website: files.eric.ed.gov/fulltext/ED604299.pdf

I pioneered Income Share Agreements (ISAs) and invented the concept of Extended Income Based Repayment (EIBR) to help cover the gap in higher education finance for students ineligible for Parent Plus and private student loans. ISAs and EIBR provide downside protection for students who graduate with debt and find it difficult to pay back their tuition. I structured the first ever financing / securitization deal for ISAs and spoke with policymakers at the White House and Capitol Hill. Our work was featured in the following research paper by Jobs for the Future and continues to inspire policymakers and higher education institutions.

I built an LLM chat application called Maya to help humans learn and discover new perspectives. Maya uses a Gemini 2.0 Flash for primary response, an uncensored distillation of R1 as a contrarian perspective, and o3-mini as a holistic perspective. Maya also present web links to go deeper into the topics discussed. The app was built in React Native / Typescript with a FastAPI / Python backend to interface with models running in the cloud and on a locally hosted ubuntu desktop with an RTX 4090 GPU.

Mediabook is an active-learning, digital textbook leveraging YouTube, Replit, and checks for understanding. I designed, built frontend, and produced video content for Mediabook starting with a project-based, intro to Computer Science course. The course was built in React / Javascript, Gatsby, and Ruby on Rails leveraging markdown / MDX as the content source (this website is built on similar technology). Video content was produced on Black Magic Pocket Cinema Cameras, DaVinci Resolve, and a Black Magic ATEM switcher.