My Work

My mission is to leverage my talents and interests to positively impact in the world. I see machine learning as a tool to maximize my impact.

At am currently Lead ML Scientist at Dendra Systems, where I work on scaling up ecosystem monitoring and restoration using smart drones.

For a 1-page summary of my career, please refer to my resume:

For a comprehensive summary, please refer to my CV:

Download ML CV (pdf)

In a past life, I was a molecular biologist. Over the years I researched colon cancer, Parkinson’s disease, olfaction, and organic chemistry.


Projects

Tackling Climate Change Using Automated Afforestation

Dendra Systems - Feb. 2020 → Now

I am currently the Lead ML engineer at a start-up fighting climate change and deforestation using tree-planting drones. Please refer to CV for detail.

Improving Remote Monitoring of Cardiac Patients by Automatically Detecting Arrythmias

Pacemate - Feb. 2019 → Dec. 2019

I worked as the lead data scientist at an implantable heart device analytics company. I implemented a pipeline to process incoming ECG transmissions from remote devices and classify arrythmias using a deep convolutional neural network. To train this network, I mined years of ground truth data labeled by board-certified electrophysiologists.

Reinforcement Learning for Tracing Neuron Morphologies

Hanchuan Peng lab. Allen Institute for Brain Science, Seattle, WA, USA - Summer 2018

I made a proof of concept showing that deep reinforcement learning can be used to automatically trace neuron morphologies.

Mosquito thermal plume tracking

Adrienne Fairhall lab. University of Washington, Dept. of Physiology and Biophysics, Seattle, WA, USA - 2014 → 2016

I worked as a research assistant in Fairhall Lab. We’re interested in answering how the history of sensory experience affects decision-making. As a model, we use mosquitoes host-seeking behavior, as carbon dioxide and thermal signals are known to inform their search. These thermal and olfactory cues have a complicated and sparse structure due to turbulence. How do mosquitoes use spatially- and temporally-sparse information to navigate to hosts? With the Daniel, Riffell and Dickinson labs, we study the behavioral and neurophysiological responses of mosquitoes to temporal patterns in heat and CO2 plumes, which we then use to develop models for multisensory integration.

My current role in the project is creating models of mosquitoes tracking turbulent thermal plumes and benchmark those models against wind-tunnel behavioral data. In particular, I am creating driven, damped agent-based dynamical models which reproduce mosquito baseline flight and then test various strategies for plume navigation. These models will be further validated using electrophysiological and tethered flight data to test theories about multi-modal decision-making.

Humpback Whale Census

Kimberley Community Whale Research Project, Murdudun, Australia - 2012

A community-initiated peer-review at the proposed site of the world’s second-largest liquefied gas processing port. Our peer-review’s estimates of humpback migration and breeding activity near James Price Point revealed gross discrepancies in the original oil conglomerate’s survey.

Mitochondrial dysfunction in Parkinson’s Disease

Leo Pallanck lab. University of Washington, Dept. of Genome Sciences, Seattle, WA, USA - 2010

I helped establish a method to grow, stain, and image primary dopaminergic neural culture from Drosophila embryos in order to test whether Parkin and PINK1, proteins involved in Parkinson’s disease, are recruited to depolarized mitochondria in dopaminergic neurons. This research was published in PNAS. I presented a poster at SACNAS 2010.

Probing Function of Colon Cancer Proteins

Brooke M. McCartney lab. Carnegie Mellon University , Dept. of Biological Sciences, Pittsburg, PA, USA - 2009

I determined that APC2, a protein with probable roles in colon cancer tumorogenesis, did not interact with β-catenin of the Wnt pathway’s destruction complex. I determined that APC2’s conserved N-terminal domain was not essential for its proper localization. This research was published in Genetics. I presented a poster at ABRCMS 2009 and Sigma Xi 2009.

Quantifying endogenous sRNA concentrations in C. elegans

Katherine Walstrom lab. New College of Florida, Dept. of Natural Sciences, Sarasota, FL, USA - 2010 → 2011

My thesis proposes a model for RNA Helicase A function in endogenous C. elegans RNAi pathways.

Searching for P450 Active site mimics

Paul Scudder lab. New College of Florida, Dept. of Natural Sciences, Sarasota, FL, USA - 2008 → 2009

I partially synthesized precursors to a novel high-valent iron-stabilizing macrocycle based on the active site of cytochrome P450.