Brigham Young University

Software Engineer at Google

Technical lead working across teams to design and implement solutions with a focus on AI efficiency and ML researcher dev velocity.

Google

Senior Software Engineer - CoreML

Building internal tools to improve Python usability and ML Research dev velocity.

Previously on the Google Compute Engine for Google Cloud. Building and maintaining data pipelines for logging various data/metrics.

Software Engineer

Worked closely on the handwriting recognition pipeline for historical documents, handwritten in various languages from around the world. Experimented with various neural network architectures to improve both the image to text recognition and the post-processing NLP.

FamilySearch

Research Engineer

Built recommendation engine for 5+ million users using Airflow, Databricks, PySpark, AWS Lambdas, AWS S3, and other tools.

Also working to improve search functionality within the problem using pairwise preference learning and LamdaMART for Information Retrieval.

Lucid

Data Scientist - Applied Machine Learning

Client facing, full-stack engineer that worked with Fortune 10 companies to address their needs and implement code in production to accommodate them. This ranged from data pipelining to user-interface changes. Enjoyed working with Qualtrics, but left to follow a career in Data Science.

Qualtrics

I worked with a team of software engineers to develop techniques for manipulating, filtering, and analyzing satellite imagery. This included developing cloud-masking algorithms, predictive models, and classifiers based on pixel values of the satellite imagery as well as other data. 

Used deep reinforcement learning to train artificial intelligent agents to play Atari games and navigate through generated mazes.

My other projects with Applied Invention include using bipartite matching to organize and find trends in large sets of data.

Applied Invention

Data Scientist

Developed a software framework to automate the Manifold Boundary Approximation Method. This software uses HTML/CSS/JavaScript as the UI, Python 3 for parsing, manipulating, and learning from data, Julia for heavy mathematical computations, and MongoDB for storage. When complete, this software should not only be able to automate a very tedious and difficult process, but it will be adaptable for various model types.

Some other research projects I have been involved with includes working on a team to build system vulnerability software for the Department of Defense using Python and HTML/CSS/JavaScript, and building chemical engineering applications for model predictive control.

Research Assistant

While awaiting admittance into a graduate program, I built a cluster of SIP socket servers to automate the phone testing process using Shell, Cron, and Python. This allowed for a testing frequency of 2-3 minutes for both domestic and international phone lines. These tests produced live reports of failure/success in a communication-heavy industry and proved to be a reliable source for the DevOps team.

I also managed the testing database for QA Automation team, which used InfluxDB for live reporting and MySQL to back up the results.  My other accomplishments at InsideSales.com include creating automated user experience tests and sampling new technologies to find potential for future products.

InsideSales.com

Quality Engineer - Automation

As an undergraduate majoring in Genetics, I found a passion for statistical models. This led me to research in the Physics Department studying statistical models of cell biology. I studied the methodology behind the Manifold Boundary Approximation Method and applied it to a complex cellular system. This system, the Wnt-pathway, was originally defined with 26 parameters and 15 differential equations. After the process of approximation, the new system (with 6 parameters and 2 differential equations), captured the same dynamics as the original system in regard to beta-catenin fold change. 
While working with MBAM, I used  Python and PyQt to develop software which improved the reduction process.

Dane Bjork