James Pettit
<james.l.pettit@gmail.com>
https://github.com/etherealmachine

Full stack application developer. Loves agile web development, building low-latency rock-solid backends, and the occasional foray into machine learning.

Experience

Software Engineer @ Google, Inc. 2014 - 2017

On the Skybox/Terra Bella team responsible for delivering satellite imagery and analytics to customers.

• Customizable web application to view, track, and analyze change over time across the globe.
• Golang backend, mixed AppEngine and microservice architecture.
• Polymer no-framework frontend.
• Personally responsible for the Spanner-based database microservice.

Technical Solutions Engineer @ Google, Inc. 2012 - 2014

On the team responsible for scaling Google support for millions of customers. Systems include:

• Web-based agent-facing customer support/relations portals (i.e. CRMs). Primarily written using an AngularJS frontend and Go-based backend with NoSQL-like databases.
• PubSub based automated processing related to order management, automated robot-customer interactions, and batch processing jobs.
• Internal auditing, authentication, and authorization of customer support requests.

Lead Programmer @ LEEPS Lab 2007 - 2012

• Published two Open Source frameworks for Experimental Economics
• Pettit, Friedman, Kephart, and Oprea. "Software for continuous game experiments". Experimental Economics Vol. 17, Issue 4.
  • A Java-based framework with a plugin architecture based on dynamic class loading.
  • Allowed rapid prototyping and completion of experiments running under soft real-time constraints in a multi-user networked environment.
  • Full PDF
  • Source and documentation available at https://leeps.ucsc.edu/cong
• https://github.com/RedwoodAdmin/RedwoodFramework
  • Next-generation experimental software development system.
  • Web-based frontend allowing HTML5 web development knowledge to be used to develop experimental software.
  • Soft real-time backend written in Go using a Redis datastore.

M.S. Computer Science, UC Santa Cruz 2011

• Focus on machine learning and artificial intelligence.
• Developed advancement to Monte-Carlo Tree Search using Evolutionary Algorithms to bias the Monte Carlo simulation.
• Applied to the game of Hex and published:
  • Pettit, Helmbold, "Evolutionary learning of policies for MCTS simulations", FDG 2012: Proceedings of the International Conference on the Foundations of Digital Games Pages 212-219
  • Full PDF
  • https://github.com/etherealmachine/hivemind

B.S. Computer Science, UC Santa Cruz 2007

Personal Projects

• https://github.com/etherealmachine/etherealmachine.github.io
  • Static site generator for this page. Built in Go with a custom Markdown parser.
  • Go, Markdown, HTTP
• https://github.com/etherealmachine/zillow-kaggle
  • Tensorflow and LightGBM machine learning models to improve the Zillow ZEstimate.
  • Python, Deep Learning, Tensorflow
• https://github.com/Leeps-Lab/otree-redwood
  • Leverage Websockets to build new experiments in the field of Experimental Economics.
  • Python, Websockets, HTML5, Javascript, Polymer
• https://github.com/etherealmachine/novaterra
  • Use Generative Adversarial Models to create terrain maps from Space Shuttle data.
  • Python, Deep Learning, Tensorflow
• https://github.com/etherealmachine/spyfall
  • Pass-and-Play the game Spyfall on you phone, pure HTML and Javascript.
  • HTML, CSS, Javascript