Master of Data Science @ University of Melbourne Β· B.Sc. in Applied Statistics @ Western University π¨π¦
- π Currently: building an internal operations platform (Next.js / TypeScript / Supabase) & finishing my data science capstone
- π± Interests: machine learning, data engineering, full-stack development, cloud & distributed systems
- π« Reach me: xfu84@uwo.ca
- π Portfolio: michaelfu627.github.io
LeNet-5 from Scratch in Pure NumPy A convolutional neural network built with no deep-learning frameworks β hand-implemented forward/backward passes, an im2col + BLAS optimization for an 11Γ per-step speedup, reaching 96.3% on MNIST, with an interactive 3D Three.js visualization of the network.
AusFlux β Cloud-Native Streaming Analytics (team project) Serverless data pipeline on Kubernetes (Fission + Elasticsearch) ingesting real-time social & weather data, deployed on the Melbourne Research Cloud.