Dave DeBarr

Dave DeBarr has more than 18 years of experience using machine learning to implement high-performance solutions for a variety of commercial and government applications. DeBarr is a principal applied researcher at Microsoft, currently focusing on authentication and account security. In his time at Microsoft, he has constructed models for user, device and network reputation; expression and activity recognition; sentiment classification; click-through rate estimation; keyword and URL recommendation for behavioral targeting segments; and sensitivity analysis for customer satisfaction. Previously, DeBarr worked as a principal computer scientist at MITRE, a federally funded research and development center. At MITRE, he worked on object detection and classification for unmanned aerial vehicles, assessment of treatment effect using doubly robust estimation, abusive tax avoidance transaction detection, and intrusion detection. He has a Ph.D. in computer science from George Mason University, where he introduced enhanced machine learning methods and representations to improve performance for computer security applications.

Experience

Principal Machine Learning Scientist

• Company Name Microsoft
• Dates EmployedApr 2014 – Present
• Employment Duration5 yrs 2 mos
• LocationRedmond, Washington
• Using machine learning to enhance customer experience and protect customers

Instructor, Machine Learning Certificate Program * * Company Name University of Washington Professional & Continuing Education * Dates Employed2015 – Present * Employment Duration4 yrs

Data Scientist

• Company Name Skytree - The Machine Learning Company®
• Dates EmployedDec 2013 – Apr 2014

• Employment Duration5 mos

  • Construction and evaluation of machine learning models for a variety of application domains
  • Sales engineering

Principal Computer Scientist

• Company Name MITRE
• Dates EmployedSep 2012 – Dec 2013

• Employment Duration1 yr 4 mos

  • Estimation of treatment effect from observational data: Gradient Boosting Machines used to produce propensity and outcome models for doubly robust estimates
  • Enhanced representation for geographical area classification using Random Forests
  • Object detection/classification for video captured by Unmanned Aerial Vehicles (... See more

Applied Researcher

• Company Name Microsoft
• Dates EmployedMar 2007 – Sep 2012
• Employment Duration5 yrs 7 mos
• LocationRedmond, Washington

Software Design Engineer, Startup Business Group

• Feature engineering and model construction/evaluation for computer vision

Bing Data Mining... See more

Artificial Intelligence Engineer, Lead

• Company Name MITRE
• Dates Employed2001 – 2007
• Employment Duration6 yrs
• Abusive Tax Avoidance Transaction Detection

Co-Principal Investigator for research and development project for scalable relational data mining technology Text mining for document ranking

Software Development Engineer, Lead

• Company Namedigi Mine
• Dates Employed2000 – 2001
• Employment Duration1 yr

Multi-Dimensional Scaling for visualizing user segments (clusters) Generation of metadata for hierarchical categorization of web pages

Education

GMU Degree NamePhD Field Of StudyComputer Science Enhanced Representation and Methods for Adversarial Learning Environments

GMU GMU Degree NameMS Field Of StudyComputer Science Semi-Supervised Machine Learning

University of Maryland Degree NameBS Field Of StudyComputer and Information Science Cumulative Undergraduate GPA: 4.0/4.0