Nathan Glenn

Flexibility, maintainability, open source collaboration

日本語

GitHub ID: garfieldnate
nathan.g.glenn@gmail.com


Skills

  1. Updating Legacy Code
  2. Machine Learning
  3. Language/Speech Processing
  4. Java
  5. Perl
  6. MySQL
  7. Git/SVN
  8. HTML/JavaScript/CSS
  9. C++

Projects


Selected Work Experience

Trivago GmbH | December 2016-Present
Development of core search product, focusing on implementing solutions for gaining greater market share. Web tools for comparison of different marketing algorithms.
Tools Java (Tomcat), Python (Pandas, PyPy, Django), Kafka, Redis, Swagger

ATR-Trek, Inc. | April 2015-October 2016
Development of an NLU system for multiple locales, including contracting and managing corpus and domain dictionary work. Development of a web app for managing client linguistic resources. Turkish-Japanese translation system, including construction of corpora and custom morphological processors. Development of utilities for machine translation such as for word segmentation, corpus cleansing, and perplexity filtering. Updating, testing and automating legacy code.
Tools Java SE/EE with Tomcat, Perl, C#, C++, Stanford Core NLP Pipeline, TATR

Soar Technology | June–August, 2012
Combining spoken user input with gestures drawn on a map to direct the path of a robot (or simulated helicopter) safely through enemy terrain. This required combining speech recognition and grammar-based semantic parsing to produce grounded instructions for the agent. Writing Android application code and integrating a C++ library for speech recognition. Developing utilities to port proprietary grammar formats to an open format used by the new system.
Tools Java, JSGF, Android, Perl, Sphinx III, Sphinx 4, Soar

Center for Language and Speech Processing, JHU | June–August, 2011
NSF funded research investigating the use of a perceptron trained on synthetic confusions for discriminative language models in speech recognition and machine translation. Experimentation and feature exploration for automatically balancing generative and discriminative model scores.
Tools Java, Perl, OpenFST, Weka

Pedagogical Software and Speech Technology Group | 2010–2012
Development and maintenance of a web application for testing student language abilities, as well as a separate research system for grading student responses and predicting language abilities automatically using automatic speech recognition and machine learning. MySql database management.
Tools Perl, Java, Sphinx 4, MySQL, PHP


Education

Linguistics BA (CS minor), Brigham Young University, Provo, UT, USA, December, 2012.
Honors Thesis: JXNL-Soar: Minimalist Parsing of Japanese in a Cognitive Modeling System.