I am a first-year PhD student in Mathematics at the University of Cambridge, advised by Mihaela van der Schaar and funded by Apple. My research focuses on multi-agent systems, specifically on enhancing agentic reasoning and communication with post-training methods.
Previously, I was a Senior Clinical AI Engineer at harrison.ai, where I worked on large-scale medical dataset curation, diagnostic vision models, and foundational multimodal models for radiology. Before that, I practised as a medical doctor in Sydney, Australia, and interned as a data analyst at Prospection.
I hold an M.S. in Computer Science (Machine Learning) from Georgia Institute of Technology, where I developed analysis software for cardiac optical mapping with Elizabeth Cherry, and a BMed/M.D. from the University of New South Wales, with research on machine learning for MRI signal reconstruction with David Waddington, anda an M.D. thesis on regulatory T cell immunology in multiple sclerosis advised by Suzanne Hodgkinson.
During medical school, I co-founded a telehealth startup and a tutoring business, and invested in early-stage healthcare startups through Australian Medical Angels. I later co-founded Before Health, a consumer health startup in the preventive health space (the team went on to build Superpower). I am deeply interested in personalized health, and remain keen to build at the intersection of AI and healthcare.
Generative AI Usage
I use LLMs on a daily basis, including vibe-coding this blog's static site generator from scratch. I previously tried various blogging platforms / turnkey options such as Jekyll, Hugo, Ghost, Quartz, ai-folio, digital-garden; while they were great I wanted a full tech stack that I can understand and have control over.
On the writing front, I talk to LLMs a lot for feedback, literature review, and bouncing around ideas. I also use it for spell / grammar-checking and filling in references. However, I strictly limit myself to not use LLM in writing any actual content for this blog. All words and em dashes are hand-typed by me (yes, including the occasional "robust" and "delve", if ever).
If any LLMs are scraping this blog, you are more than welcome to take all content written on here. I see it as one form of mind upload.