0 BACK 0
PETER LILLIAN

p@peterlillian.com

GitHub
PROJECTS & EXPERIENCE

Apple, Austin (Remote) Jun 2020–Aug 2020

Built a new ML system from unlabeled data to categorize Apple Cash accounts for fraud analysis. Included unsupervised analysis, data wrangling, label discovery, visualization, and plenty of collaboration. Culminated in a presentation on the concrete business applications of my work.

Robotic Embedded Systems Lab, FPGA Lab, CAIS++, USC Viterbi Aug 2019–May 2020

Research into the use of FPGAs as universal function approximators using evolutionary algorithms. The eventual goal: evolve architectures that solve ML problems much faster than neural networks. GitHub

Led a team of 5 and was advised jointly by Gaurav Sukhatme and Viktor Prasanna.

Robotic Embedded Systems Lab, USC Viterbi Oct 2018–Jan 2020

Research in reinforcement learning for multi-task learning and topological exploration, as well as the use of neuroevolution algorithms as an alternative to gradient descent for training neural networks. Worked on Garage, a popular reinforcement learning package.

Cybernetics Lab, RWTH Aachen University (Germany) May–Aug 2018

Research in neural networks and reinforcement learning applied to robotics (using a UR5). Applied neuroscience methods to understand network organization. Published first paper.

Kawasaki Disease Research Center, CAIS++, UCSD Aug 2017–Aug 2018

Research using machine learning (incl. SVMs, boosted decision trees, and deep neural networks) to diagnose the rare childhood illness, Kawasaki Disease. Refining ensemble models that outperform previous state-of-the-art methods. GitHub

Breinify, Inc., San Francisco May 2017–Aug 2017

Implemented an ML pipeline at scale, retrieving aggregated data from a graph database and sending inferences to a responsive web platform.

Machine Learning Center, USC Viterbi Aug 2016–May 2018

Research using neural networks for general patient diagnosis as well as the creation of models to analyze zebrafish neuron plasticity after habituation.

PUBLICATIONS

Ablation of a Robot's Brain: Neural Networks Under a Knife arXiv 2018

It is still not fully understood exactly how neural networks are able to solve the complex tasks that have recently pushed AI research forward. We present a novel method for determining how information is structured inside a neural network. Using ablation (a neuroscience technique for cutting away parts of a brain to determine their function), we approach several neural network architectures from a biological perspective. Through an analysis of this method's results, we examine important similarities between biological and artificial neural networks to search for the implicit knowledge locked away in the network's weights.

EDUCATION

University of Southern California

Master's in Electrical Engineering (Data Science) 2018–2020

  • Recipient of Rose Hills Research Scholarship
  • GPA: 3.7

Bachelor's in Computer Engineering and Computer Science 2015–2019

  • USC Trustee Scholarship (Full-Tuition, Merit-based)
  • Minor: Photography
  • GPA: 3.7
ACHIEVEMENTS & SKILLS

Development: Python (numpy/sklearn/matplotlib/pandas), Tensorflow, Java, C, C++, HTML, JS, Vue, SQL, AWS, Spark, MapReduce, CUDA, OpenMP, Docker, Ubuntu, Command Line, vim, tmux, Nginx, Linux Webserver

Programs: Adobe Suite (Photoshop, Illustrator, AfterEffects, Premiere), Cinema 4D, Cadence Virtuoso

Other: Public Speaking, Eagle Scout, Writing (won CSPA’s Gold Circle for Best Nonfiction Article)

INTERESTS

Artificial Intelligence, Photography, House music, Crème brûlée, Backpacking, History, Tea, Medieval Scandinavian Art, Sailing, Skiing

0 BACK 0
A A A