LinkedIn

Senior Staff Software Engineer, Machine Learning

Tech lead of the Anti-Abuse AI Team (>40 ML engineers and managers).

I've led the technical development of advanced modeling and signals capabilities including:

■ Detection of Generative AI content: Created the first deepfake video detection system at LinkedIn using deep learning. Led a team that created the first AI-generated photo detection model at LinkedIn.
■ Pioneered the use of deep learning to detect abusive automated sequences of user activity--e.g. using natural language processing (NLP) techniques to classify sequences of user actions instead of sentences of words

AI engineer productivity:

■ Led a two-year, 20-engineer cross-functional effort to up-level Anti-Abuse AI developer productivity.

Staff Software Engineer, Machine Learning

Tech lead of the anti-scraping and automation AI team (i.e., detecting and preventing automated bot behavior).

I proposed the use of unsupervised outlier detection for scraping and bot detection, created a Spark/Scala implementation of the isolation forest algorithm, open sourced it, brought the first application using this library to production, and subsequently led a team of ML engineers that leveraged my library and methodology in more than five separate production use cases.

Senior Software Engineer, Machine Learning

I built abusive entity reputation and fake account detection machine learning models while working in the Anti-Abuse AI Team.

Software Engineer - Data Mining/Data Analysis/Machine Learning

Insight Data Science

■ Created an account takeover (ATO) detection system while consulting for Castle (YC W16)
■ Developed a statistical model for unsupervised anomalous event detection resulting in a >2× improvement in the recall of compromised accounts over the existing model with the same false positive rate

Data Science Fellow

Brown University

■ Led the experimental design, operations, and data analysis effort for a novel, world-leading calibration of the world's most sensitive dark matter detector
■ Created, deployed, and operated a Python-based data processing framework on an 8,000 core high-performance computing cluster processing up to 1 TB/day of incoming data
■ Architected and administered the primary data repository and distribution system serving 500 TB to 100+ worldwide collaborators
■ Managed the data processing working group while scaling up the data pipeline for production use
■ Implemented algorithms to characterize features (10 ns to 300 μs) in time series data
■ Pioneered the use of SciDB using AWS for interactive analysis on TB-scale datasets
■ Led development of the Python-based data processing framework supporting Python, Matlab, and C++/ROOT algorithm modules
■ Directed multiple teams during 307 days of on-site presence during the integration and operation of the LUX experiment 4,850 ft underground at Sanford Laboratory in Lead, SD
■ Managed a team of scientists and engineers as a calibration working group leader

Particle Astrophysics Group

I use AI to detect and prevent abuse at scale.

Taplio

James Verbus's Linkedin Analytics

James Verbus

Want to drive more opportunities from LinkedIn?

Taplio

.css-1jgreyo{background-image:linear-gradient(284deg, rgba(39,43,72,1) 0%, rgba(33,47,183,1) 100%);color:transparent;-webkit-background-clip:text;background-clip:text;}James Verbus's Linkedin Analytics

James Verbus

Want to drive more opportunities from LinkedIn?

James Verbus's Linkedin Analytics