Ryan Peterz

Digital Nomad · (000) 000-0000 · ryan.peterz@gmail.com

I am interested in opportunities to advance AI Safety research

I started programming and puberty at the same time



Resume

Skills

Fullstack

Databases {SQL: SQLAlchemy, noSQL: mongodb}

Frameworks [pytest, FastAPI]

Languages {python: 3.9, JavaScript: [react, d3], Java: Android, C#: Unity}

Machine Learning

Models [Reinforcement Learning, Image Processing]

Frameworks [pytorch, tensorflow, keras, ray]

Experience

2018 - Current
Data Scientist & Machine Learning Engineer (Geometric Data Analytics, Inc.)

Developed and used machine learning software to deploy and reproducibly run large experimental computations, including unsupervised annotation of ultra-high resolution images in the RootShape project, recursive multi-agent reinforcement learning algorithms (for DARPA), and persistent homology calculations across an experimental matrix of neural network subjects to empirically study a theoretical generalization bound (for NGA).

2016 - 2018
Teaching Assistant, Mathematics (University of Kentucky)

Instructed undergraduates in Calculus I & II and Finite Mathematics, a course incorporating concepts from combinatorics, finite probability, linear algebra and convex optimization. Student rosters ranged roughly between 60~120 per semester.

Summer 2013
Data Analysis Intern (SACHEM, Inc.)

Aggregated chemical manufacturing data in various formats (.csv, .xls, .rpt) by developing automated parsing methods adaptive to data storage standards that varied between departments and over time. Analyzed data using python data science tools (pandas, scikit) to identify periodic and outlying features. Translated academic paper into ARIMA predictive model. Scheduled automatic data aggregation and batch analysis using VBScript.

2007 - 2012
Warcraft III Modding (Hobby)

Progressed from the game editor's GUI to Blizzard's JASS (just another scripting system; a verbose, imperative, event-based scripting language with an incredibly useful bug) to the community-hacked preprocessor vJass. Explored physics, geometry and digital art through the imagination and implementation of original magical spells. Tackled UI and UX by hacking the finite item interface to simulate an infinite inventory. Programmed procedural AI to collect resources and attack the enemy. Simulated an island ecosystem with plants, herbivores and carnivores.

Education

2016 - 2018
M.A. Mathematics University of Kentucky

Composed and presented "Introduction to Neural Networks" at the Graduate Student Colloquium

Investigated the effects of a clever orthogonalization scheme on very deep convolutional neural networks (slides).

Implemented a recursive decomposition for reducing terms in the algebra of Steenrod Squares (github)

2012 - 2016
B.S. Mathematics University of Kentucky

Cofounder and President of the U.K. Consciousness Club

August 2012 - May 2018

Loosely Affiliated Decentralized Dude Society

TJ Keel

Lihan Yao

Thomas Barron

Steve