I am a machine learning engineer with a focus on research. I have extensive experience in applied ML to signals, and biomolecule sequences.
Here is a list of (some of my) previous achievements:
- Researched, designed, developed and optimised deep learning on time-series signals. Analysed 500,000 patient's health data to predict metabolic diseases from physical activity accelerometer sensors. Leveraged transfer learning to prototype quickly. Architectured a novel predictive model called 'SpecNet' to narrowly optimise predictions on spectrograms in TensorFlow, increasing test accuracy by 18%. First-class awarded.
- Developed an ensemble CNN architecture using majority voting and optimised pyramidal kernel structure to achieve state-of-the-art accuracy on the MNIST dataset (0.9934), leveraged vectorised computing SIMD units to speed up training on a HPC system with over 5,000 CPUs.
- Built a reinforcement learning agent in a discrete, dynamic, stochastic environment to outperform human performance using a Marcovian Decision Process via value iteration through the Bellman equation.
- Optimised and implemented an intelligent computer vision edge detection algorithm, that matches human behaviour and interpretation of such edge discontinuities with an excellent average f1 score of 0.71 on unseen images. (83% awarded).
