latest news!
My paper “Evolution and learning in differentiable robots” has been accepted to Robotics: Science and Systems 2024. Herein we report using differentiable physics simulation to conduct the largest evolutionary robotics experiments ever reported. Learn more and see our submission to the ALife 2024 Virtual Creatures Competition here.
about
I am a PhD student in the xenobot lab inside the Center for Robotics and Biosystems at Northwestern University. My research program focuses on brain-body co-design in robots and experiments in large-scale artificial life.
In the private sector I sit on the board of and consult with Vaginal Biome Sciences, a biotechnology company developing novel microbiota-based therapeutics.
past work
I earned my MSc. in Computer Science at the University of Texas at Austin in 2023. During this time I worked with David Harwath in the SALT Lab and developed the leading methodology for supervised and unsupervised phonetic speech segmentation. My thesis can be found here.
Before that, in the QBI Lab at Oregon Health and Science University, I developed an evaluation framework for unlabeled cell segmentation, and in the Gardner Lab at the University of Oregon, I designed a neural model for multi-scale spectral analysis of bird song.
Outside of academic research, I have actively contributed to the real-time gesture recognition systems being built by Skywalk Inc. I have also worked as a software engineer, data scientist, and technical consultant for Fortune 500 and Y-Combinator companies.
I received my B.A. in Computer Science at UC Berkeley in 2018 where I derived a mathematical model describing the effect of environmental pollution of parasitic disease transmission.
publications
contact
You can reach me via direct message on twitter.
Curriculum Vitae
Education
PhD, Computer Science, Northwestern University
2023 - 2027 (expected)
MSc, Computer Science, University of Texas at Austin
2021 - 2023
BA, Computer Science, University of California at Berkeley
2014 - 2018
Experience
Center for Robotics & Biosystems, Northwestern University - Graduate research staff
Chicago, IL. 2023 - Present
Created and open-sourced a GPU-accelerated robot simulator and used it to conduct the largest-scale evolutionary robotics experiments ever reported.
Developing methods for learning adaptive and goal-directed robot control using differentiable physics simulation.
Employing soft-body simulation techniques (particle and mass-spring) to co-optimize robot brain-body systems using evolutionary and gradient-based methods.
Member Board of Directors, Vaginal Biome Sciences
Portland, OR. 2022 - Present
Assisting company executives with strategic planning, grant application writing, and pitch deck development.
Meet quarterly with company Board of Directors to approve large expenditures and employee compensation packages.
Graduate Research Assistant, Speech Audio & Language Technologies Lab, University of Texas at Austin
Austin, TX. 2021 - 2023
Created and open-sourced the state-of-the-art methodology for supervised and unsupervised speech segmentation.
Employed large-scale contrastive pre-training to learn effective data representations and enable highly data-efficient downstream fine-tuning.
AI Research Engineer, Skywalk LLC
Palo Alto, CA. 2022 - 2023
Built company’s core ML codebase including data loading and normalization, end-to-end training, and real-time inference.
Designed and implemented neural network architectures for embedded, gesture recognition from wrist-based physiological data readings.
Research Engineer, Quantitative Bio-Imaging Lab, Oregon Health and Science University
Portland, OR. 2021 - 2022
Developed and open-sourced automated tool for evaluating cell segmentation algorithms without ground-truth labels.
Designed and trained variational autoencoders to represent, cluster, and classify large-scale human cell and tissue imaging data.
Research Assistant, Gardner Lab, University of Oregon
Eugene, OR. 2021 - 2022
Designed and trained convolutional and recurrent neural networks for automated segmentation of birdsong audio.
Developed software pipeline for automated audio acquisition and spectral feature processing of birdsong.
Independent Contract Work — SWE, Data Scientist, Consultant
Portland, OR. 2020 - 2021
Work for Nike and Mighty Computing
Publications
Skills and tools
Highly proficient in:
Python development
PyTorch neural-network development and training
Matplotlib & seaborn data visualization
NumPy data analysis
Taichi differentiable, gpu-accelerated simulation
Git-based version control
Past experience with:
C++ development
CUDA development
JAX neural network development and training
scikit-image biological image analysis
Adobe Premiere Pro video editing
Adobe InDesign layout designs
Adobe Illustrator vector graphics
Graduate coursework
Deep learning
Swarm robotics
Machine learning
Natural language processing
Reinforcement learning
Artificial life
Computer vision
Linear algebra
Parallel systems
Computer networking