Graham Neubig

Graham Neubig

About

Detail

Co-Founder
Pittsburgh, Pennsylvania, United States

Timeline


work
Job
school
Education
auto_stories
Publication

Résumé


Jobs verified_user 0% verified
  • OpenHands
    Co-Founder
    OpenHands
    Jun 2024 - Current (2 years 2 months)
  • I
    Co-Founder and CEO
    Inspired Cognition
    Sep 2021 - Oct 2023 (2 years 2 months)
  • Carnegie Mellon University
    Associate Professor
    Carnegie Mellon University
    Jul 2020 - Current (6 years 1 month)
  • Carnegie Mellon University
    Assistant Professor
    Carnegie Mellon University
    Aug 2016 - Jun 2020 (3 years 11 months)
  • H
    Coordinator For International Relations
    Hyogo Prefectural Government
    Aug 2006 - Mar 2008 (1 year 8 months)
  • Nara Institute of Science and Technology
    Assistant Professor
    Nara Institute of Science and Technology
    Apr 2012 - Jul 2016 (4 years 4 months)
  • T
    Assistant Language Teacher
    Tajima Agricultural High School
    Aug 2005 - Jul 2006 (1 year)
Education verified_user 0% verified
  • Kyoto University
    Doctor of Philosophy - PhD, Informatics
    Kyoto University
    Jan 2010 - Dec 2012 (3 years)
  • Kyoto University
    Master's degree, Informatics
    Kyoto University
    Jan 2008 - Dec 2010 (3 years)
  • University of Illinois UrbanaChampaign
    Bachelor's degree, Computer Science
    University of Illinois UrbanaChampaign
    Jan 2001 - Dec 2005 (5 years)
Projects (professional or personal) verified_user 0% verified
  • A
    All Hands AI
    Collaborated with researchers at OpenHands (https://www.linkedin.com/company/openhands-ai/) , an open-source initiative that aims to build a platform for autonomous software development. (Advised by Graham Neubig) A major challenge with the CodeAct Agent (the foundation of OpenHands) was its poor file editing capability. The SWE-bench is a benchmark designed to evaluate LLMs on real-world software engineering problems. Evaluating on the SWE-bench is a resource-intensive and time-consuming process. To solve the problem of effectively testing iterative improvements, I integrated the Aider Editing Benchmark as an alternative to evaluating the agent’s file editing abilities. Additionally, I implemented visualizations to track the agent's perfor
  • M
    Multilingual Models for Neural Machine Translation
Publications verified_user 0% verified
  • N
    When and Why are Pre-trained Word Embeddings Useful for Neural Machine Translation?
    NAACLHLT Proceedings of the Conference of the North American Chapter of the Association for Computational Linguistics Human Language Technologies
    Jun 2018
This is a community-created genome.