Phillip Taylor

Princeton University

• Relevant Coursework: Computer Vision, Information Security, Operating Systems, Quantum Computing, Automated Control Systems, Economics and Computation, Advanced Programming Techniques, Data Structures and Algorithms, Programming Systems, Reasoning About Computation, Computer Music

• Activities:

  • Terrace F. Club - Music Director

    Manage a yearly budget for booking entertainment at Terrace F. Club. Secure logistics, equipment, accommodations, and financial settlements to prepare weekly live shows.

  • Princeton Jazz Department - Small Group I Pianist

    Play shows and concerts at Princeton University and in the Princeton community.

  • COS429 (Computer Vision) TA

    Grade assignments, tests, and hold office hours to assist students in COS429.

  • COS217 (Programming Systems) Grader

    Grade assignments, tests, and assist students in COS217.

Princeton Visual AI Lab

Princeton, NJ

• Develop methods of video understanding using computer vision and machine learning to analyze policing footage in order to evaluate racial biases and systemic trends in policing organizations, ultimately to improve police accountability and performance.

• Read more about our research here: policingresearch.org

RestAR

Tel Aviv, Israel

• Built a tool to integrate Blender and ThreeJS into a Jupyter environment to customize, view, and interact with 3D visualizations of products.

• Used to fix autogenerated 2D texture mappings onto 3D meshes.

Qatar Computing Research Institute

Doha, Qatar

• Designed and created a multi-threaded, distributed traffic simulation using SUMO (Simulation of Urban Mobility) in C++ to enable parallelization on multiple cores.

• Implemented for fast training and testing of a reinforcement learning model to optimize traffic signal control in Qatar to compensate for the projected increase in traffic during the FIFA World Cup in 2022.

• Able to simulate realistic traffic loads on a city and country-wide scale.

• Research paper accepted into SIGSpatial conference 2020 and SUMO User Conference 2020:
  QarSUMO: A Parallel, Congestion-optimized Traffic Simulator

Sheikh Zayed Institute for Pediatric Surgical Innovation

Children's National Hospital - Washington, DC

• Developed a novel deep learning method using Python and Microsoft CNTK to detect genetic diseases in newborns and young children based on pictures of faces.

• Improved accuracy for key facial landmark estimation and automated genetic disorder screening over previous state-of-the-art methods.

Max Planck Florida Institute for Neuroscience

Jupiter, FL

• Designed, built, and programmed an inexpensive automated perfusion system using a Raspberry Pi and Python to maintain cellular life for imaging and experimentation at the microscope stage.

• Created a Python Tkinter GUI for easy researcher interface with the system.

Towards DESRED: A Diverse Explorable Super Resolution Dataset

My thesis introduces a novel problem formulation in the computer vision task of explorable super resolution - reconstructing multiple possible high-res images from a low-res input - and provides a solution for that problem.

• No proper dataset exists yet for training, testing and evaluating ESR methods.

• My thesis constructs this dataset using a novel methodology that outperforms previous ESR methods.

• Github link.

Subscrypto

Web application that implements blockchain-native payment-over-time on the Ethereum network.

• Smart contract automates payment process to ensure both parties always have access to the proper funds.

• Github link.

Princeton AdvisorSearch

• COS333 (Advanced Programming Techniques) Capstone Project.

• Enables students to search for advisors based on interests, connect with professors, and view past students' work.

Eye2Eye: Eye Tracking for Social Cue Training in Children with Autism

Eye2Eye is a VR game built for the FOVE VR headset designed to help children with autism practice making eye contact and focusing on faces.

• Utilizes the built in eye tracking capabalities of the FOVE headset to have in-game response to eye contact in real time.

• Precise data on eye movements for each game session enables provision of tailored feedback and accurate evaluation.

• Developed in C# with Unity using FOVE's Unity plugin.

Jazz/Classical Piano

Soloist, small and large ensemble player. Play concerts, gigs for events, and provide accompaniment for student performances.

Performed with Dave Holland, Ambrose Akinmusire, Danilo Perez, Ron Blake, Bobby Broom, Ugonna Okegwo, Dion Parsons.

Studied under Stephen Scott, Kris Davis, Angelica Sanchez, Alan Feinberg.

Awards and Honors:

• 2017 Berklee Jazz Festival Winner

• 2017 Jazz Band of America Pianist

• 2016 Downbeat Magazine Jazz Soloist Outstanding Performance Award Winner

• 2015 Florida All-State Jazz Band Pianist

Software Development

Programming Languages:

• Java, Python, C, C++, C#, MATLAB, Javascript, HTML+CSS, x86 Assembly

Frameworks:

• ML: PyTorch, TensorFlow, CNTK

• Graphics: three.js, Blender, Unity

• Web: Heroku, Flask, jQuery, Bootstrap

• Web3: Solidity, Truffle, Node, Web3.js