About

About

Hi, I’m Fortune Olawale

Electrical Engineering

GitHub Medium Email


me

I am an Electrical Engineering student at the University of British Columbia interested in machine learning, literature, music, and a lot of other things.

My work has taken me across several fields:

  • building 5.8 GHz radio-frequency systems for small-drone detection;
  • deploying medical-imaging models used across hospitals in Nigeria;
  • studying neural-network optimisation as a numerical ODE problem;
  • and building my own tiny protein language model.

I like solving problems.


Beyond the Code

Some of my favourite things to talk about:

  • music, instruments, and music production;
  • books, especially science fiction and mystery novels;
  • computer chips and transistor design: MOSFETs, FinFETs, and GAAFETs;

Selected Work

nano-PLM

A small protein masked-language-model project built to explore \ the complete training pipeline: residue tokenisation, sequence batching, \ bidirectional attention, and modern optimisation.

Ideas: PyTorch, protein sequences, Transformers, RoPE, and Muon.

Runge–Kutta Optimisers

An experimental comparison of AdamW, Muon + AdamW, and preconditioned RK4 on MNIST, \ viewing neural-network training through gradient flow and numerical integration.

Finding: Higher-order integration alone does not guarantee better optimisation;
the batch regime matters considerably.

Flow Matching

A compact implementation for understanding how a learned vector field can transport samples \ from a simple distribution towards a target distribution.

Focus: Learning by implementing, visualising, and testing the underlying dynamics.

SRFBN-S ×4 Super-Resolution

An implementation of a feedback network for ×4 image super-resolution, \ trained on DIV2K and evaluated on Urban100.

Lesson: Always inspect AI-generated code carefully.

ESOL Graph Neural Network

An overfitted graph neural network for predicting molecular solubility from atomic structure.

Ideas: Molecular graphs, message passing, and learned representations.

Okanagan Rover Craft Club

Currently working on machine-learning development and version control
for a student rover team preparing for the Canadian International Rover Challenge.

Systems: ROS 2, cameras, NVIDIA Jetson, and robotics.


Tools I Reach For

Languages and Machine Learning

Python PyTorch C++ R Jupyter

Engineering and Development

Linux Git GitHub ROS 2 Django

I also work with spectrum analysers, directional and omnidirectional antennas, RSSI-based ranging, and data analysis.


Writing

Read my essays and technical notes on my homepage


Away From the Terminal

You will probably find me:

  • reading a book;
  • playing guitar or bass;
  • listening to music;

Reach Me

I am always open to interesting problems, research ideas, and engineering collaborations.

Email · GitHub