Gerard Goossen
Neural Network Engineer at minds.ai- Claim this Profile
Click to upgrade to our gold package
for the full feature experience.
Topline Score
Bio
Experience
-
minds.ai
-
United States
-
Software Development
-
1 - 100 Employee
-
Neural Network Engineer
-
Oct 2022 - Present
-
-
-
Facebook
-
India
-
Banking
-
Tech Lead
-
Jan 2015 - Nov 2020
Research, design and development of large scale classification systems for various abuse detection teams. I was responsible for setting direction, technical leadership, contributing to roadmaps and visions, growing and scaling the capabilities of the team. Specific efforts are listed below. • Researched and developed machine learning ideas for high-impact improvements to abusive account detection. Identified fundamental limitations of the existing approach and resolving these by introducing using Graph Neural Nets, transfer learning and autoencoders. These resulted in large increases in recall and opened up more avenues for improvement. • Drove usage of classification scores generated by my team. The scores are used in hundreds of places and forms the basis for basically all abuse detection in Facebook. Supported teams to make optimal use of the scores and collaborated to ensure stability and adversarial robustness. • Architected solutions in collaboration with specially created infrastructure teams to migrate to a graph neural network and optimize resource usage to save thousands of servers.
-
-
-
Textkernel
-
Netherlands
-
Software Development
-
100 - 200 Employee
-
Senior Software Engineer
-
Apr 2010 - Nov 2014
Development, Research and design of Natural Language Processing models. • Extracted information from various textual content such as small text messages, web pages, and legal documents, covering 15 languages. • Created models that are complex pipelines of tasks such as sequence labelling, named entity recognition classification, and manual rules. • Development included creation of frameworks, tools, visualisations, automated tests to do feature engineering, evaluation, annotation and reasoning. • Authored a Pattern Matching Engine with a Domain Specific Language optimised to work on the word level. The engine is used in 5 different projects and has over a 1000 rules written for it.
-
-
-
The Perl Foundation
-
Software Development
-
Grant provided by the Perl foundation
-
Nov 2009 - Jul 2010
Grant: Changing the Perl 5 optree build process into an Abstract Syntax Tree generation and a code generation step. Refactoring part of the core of Perl, thereby increasing its modularity, adding just-in-time-compilation, and pave the way for further improvements. Grant: Changing the Perl 5 optree build process into an Abstract Syntax Tree generation and a code generation step. Refactoring part of the core of Perl, thereby increasing its modularity, adding just-in-time-compilation, and pave the way for further improvements.
-
-
-
-
Senior Software Engineer
-
Oct 2004 - Nov 2009
Development and maintenance of high traffic websites with extensive database integration. • Designed and developed a Search Engine consisting of 5000 lines of C++ code and indexing 3+ million documents with performance of 10,000 queries per minute. • Created and deployed a distributed file system serving 100+ million files. Development and maintenance of high traffic websites with extensive database integration. • Designed and developed a Search Engine consisting of 5000 lines of C++ code and indexing 3+ million documents with performance of 10,000 queries per minute. • Created and deployed a distributed file system serving 100+ million files.
-
-
Education
-
University of Amsterdam
M.Sc., Physics