Hi there, I'm Nick.
I'm passionate about using machine learning to solve big industry problems (energy, health, infrastructure, agriculture, education etc).
Projects
At the University of California, Berkeley, I've been working on energy disaggregation. My tool, DisCoVA, uses variational autoencoders with convolutional layers. I was inspired by their lower reconstruction loss vs. autoencoders in classic image datasets. Check out a snapshot below of some projects I've worked on.
Background
I got my B.Sc. in Discrete Mathematics and Operations Research at the University of Melbourne (with study abroad at UC Berkeley). I then worked in equity research at Goldman Sachs covering the energy and utilities sector, before returning to UC Berkeley in 2017 for an M.Eng. in IEOR.
About Me
I left my home and job and came back to California because I was itching to find a career working on hard, high impact problems. I have solid math fundamentals, strong written and verbal communication skills, commercial acumen, intense curiosity, and a desire to learn quickly from others.
DisCoVA: An energy disaggregation tool using variational autoencoders with convolutional layers.
Implemented an adversarial agent from Poisoning Attacks Against Support Vector Machines (Biggio et al.).
EmbedRank: Exploratory word importance ranking using PageRank on a similarity matrix derived from an embedded space.
The Brownlow Downlow: Predicting the vote distribution for the AFL MVP using balanced bagging.
Predicting car service operations from car and engine data with CarForce.
I've been working on some new projects and will update shortly.