Jamie Lo

Portfolio — 2026

Jamie Lo

I build machine learning models that read medical images — and small iOS games.

View projects GitHub (opens in new tab)

01

Selected work

01Machine learningPython

Retinal disease classification

Classifies retinal fundus images with three model families — a ResNet-style CNN, a vision GNN that builds dynamic k-NN graphs over image patches, and a hybrid that feeds CNN features into graph layers. Predictions are served through a FastAPI inference API and a plain HTML/CSS/JS frontend.

  • Binary disease-risk and multilabel detection of DR, MH, TSLN and ODC
  • Trained and evaluated on the RFMiD dataset with selectable GCN, GAT and GIN graph layers
  • Web demo with test-set gallery, image upload and adjustable thresholds
  • Python
  • PyTorch
  • PyTorch Geometric
  • FastAPI
Repository for Retinal disease classification (opens in new tab)

02HackathonTeam of 52026

LostPoint

An AI lecture companion built with a team of five at GreatUniHack 2026. Students mark the slide where they got lost, and a coach scoped to that lecture helps them through it, citing real slide numbers.

  • One-tap "I'm lost" moments with offline sync and duplicate merging
  • Guide, explain and hint coach modes, with assessed coursework kept locked
  • Live lecture sockets, anonymised per-slide heatmaps and tutor handoffs
  • React
  • FastAPI
  • SQLite
  • WebSockets
Repository for LostPoint (opens in new tab)

03iOS gameSwift

Bloodstream

An educational SpriteKit arcade game about atherosclerosis. You steer a red blood cell through an artery, dodging plaque that grows larger and faster over time, while info screens explain how plaque forms and how statins help.

  • Touch and swipe controls with SpriteKit physics and collision categories
  • Statin power-ups reset the dynamic difficulty ramp
  • Persistent JSON leaderboard with optional end-of-game selfies
  • Swift
  • SpriteKit
  • UIKit
Repository for Bloodstream (opens in new tab)

04Hack CampTeam of 42024

Red Ocelot

A PHP and PostgreSQL dashboard built with a team of four at Hack Camp 2024. It studies how student teams commit over the life of a project, from frequent daily pushes to weekly bursts.

  • GitHub API ingestion of repositories, branches, commits and files into PostgreSQL
  • Line, bar, pie and histogram views of commit activity
  • Interactive diagram of the database schema
  • PHP
  • PostgreSQL
  • JavaScript
  • GitHub API

Private repository

02

About

I'm Jamie, a developer interested in applying machine learning to medical imaging and in building approachable software around it.

My projects range from comparing convolutional and graph-based neural networks for retinal disease detection to a small iOS game that teaches how atherosclerosis works.

I also enjoy hackathons, where I've built tools for students and teams alongside others under tight deadlines.

Machine learning
PyTorch, PyTorch Geometric, CNNs, graph neural networks
iOS
Swift, SpriteKit, UIKit, storyboards
Web & tooling
React, FastAPI, PHP, PostgreSQL, HTML, CSS, JavaScript

03

Spare time

What I get up to when I'm not building:

  • Running

    Time outside and on my feet — the best way to reset after a long build.

  • Gaming

    Single-player and co-op games, especially ones with systems worth understanding.

  • Piano

    Playing for the joy of it, working through whatever piece I'm learning.

  • Horror

    Films, games and fiction — the slower and stranger, the better.

04

Contact

Open to conversations about projects, roles and collaborations.

[email protected]