Brendan Herger

Hey, I'm Brendan

Hey there! I'm a Machine Learning Engineer, focusing on Deep Learning, Natural Language Processing, and bringing data science into production.

In the past, I’ve founded of Capital One's Center for Machine Learning, founded and lead the Data Intelligence Conference, and advised for H2O.ai and startup.ml.

In my (little) spare time, I enjoy bike riding, baking soufflés, play video games, and traveling. My recent adventures have included Beijing, Tokyo, Paris, Mt. Fuji, Portland, LA, New Orleans, Las Vegas and quite a few other journeys!

 
 

As a founding member of Capital One’s machine learning group, I have lead machine learning efforts by scoping, implementing, and delivering projects. Responsibilities have included training models, translating business needs to project plans, providing code reviews, and guiding & coordinating team members. Projects have included:

  • Subpoena Automation: Developed a patent pending framework for automatically triaging, classifying, and extracting information from unstructured legal documents, leveraging Natural Language Processing and Deep Learning

  • Authorization Fraud: Developed tree based models to catch fraudulent credit card swipes, with an emphasis on class imbalance and identifying new attack vectors. Deployed models in a sub-millisecond streaming pipeline

  • Insider Threat: Coordinated engineering and modeling efforts to find employees acting maliciously or illegally, by building a holistic employee data lake and deploying models for employee churn and data exfiltration

I've also worked with a variety of technologies:

  • Modeling: Natural Language Processing, Deep Learning, Class Imbalance

  • Languages / Software Engineering: Python (preferred), Java (prior exp.), Scala (prior exp.), shell, SQL, GitHub

  • Frameworks: Keras (TensorFlow), Python data ecosystem (Pandas, Numpy, SK-Learn), NLTK, Hadoop ecosystem (HDFS,Yarn, Hive, Spark), AWS

  • Project Management: Certified Scrum Master (CSM), Agile, Jira