Hugh Perkins

Member of Technical Staff @ Genesis AI

About

Machine learning and NLP research: - wrote DeepCL, https://github.com/hughperkins/DeepCL - ported Torch to opencl, https://github.com/hughperkins/cltorch - wrote the very first pytorch https://github.com/hughperkins/pytorch - wrote a CUDA compiler for OpenCL 1.2 devices, https://github.com/hughperkins/coriander - "Dialog intent induction with deep multiview clustering", Hugh Perkins, Yang Yi, EMNLP 2019

Country

City

United States

Industry

Computer Software

Skill

Machine Learning, OpenCL, CUDA, C++, OpenGL, C#, Java, Python, Data Mining, Matlab, Linux, Unix, Bash, Windows, Windows Server, JavaScript, MPI, GPGPU, SWIFT, Testing

Experience

Member of Technical Staff

Genesis AI

2025-5 - Present · 1 yr 5 mos

Research Engineer

ASAPP

2019-11 - 2025-5 · 5 yrs 7 mos

New York City Metropolitan Area

Member of Research team: - NLP research - NLP product development

Machine Learning Engineer

ASAPP

2017-11 - 2019-11 · 2 yrs 1 mo

New York City Metropolitan Area

NLP product engineering. Scalability and security engineering.

Machine Learning Engineer

ASAPP

2016-1 - 2017-11 · 1 yr 11 mos

London, England, United Kingdom

NLP product engineering. Scalability and security engineering.

Test and Release Automation Engineer

Standard Chartered

2013-8 - 2015-12 · 2 yrs 5 mos

Tianjin, China

2015: Senior Devops Engineer, Financial Markets SABRE team 2014: TLM/CMS Dev Lead, Financial Markets REALM team 2013: Devops Lead, Financial Markets Platforms team

Infrastructure support, scripting and engineering

Societe Generale

2002-8 - 2008-2 · 5 yrs 7 mos

Paris, France

Infrastructure scripting and engineering in the Windows desktop and Windows server teams

Education

Tsinghua University

Computer Science

2011 - 2013 · 2 yrs

Research masters in computer science, specializing in machine learning Papers published: “Gibbs Max-Margin Topic Models with Fast Sampling Algorithms”, Jun Zhu, Ning Chen, Hugh Perkins, Bo Zhang, accepted by ICML 2013 Papers submitted: a new method to do parallel large-scale SVM training, submitted to SIGKDD 2013

Shenzhen University

Chinese (Mandarin)

2010 - 2011 · 1 yr

University of Cambridge

Natural Sciences

Year 3: Psychology (class II.1) Year 2: Advanced Physics, Experimental Psychology (class II.1) Year 1: Advanced Maths, Physics, Geology, Chemistry (class I)