Skip to main content

Putting Machine Learning to Work for Digital Forensics


Jeff LeJeune and Bazyli Debowski

Jeff LeJeune, Vice President of Engineering, is responsible for all engineering and product development activities at Magnet Forensics. Jeff has over 20 years of software industry experience starting as a software developer before moving into leadership roles. Most recently Jeff was Director of Engineering at Avast Software and prior to that he held many senior leadership positions in nearly 12 years at BlackBerry, formerly Research in Motion (RIM), Limited. Jeff holds a Bachelor of Mathematics in Computer Science degree from the University of Waterloo. Bazyli Debowski has been working at Magnet Forensics as a Software Developer on the Data Analytics team for 2 years. His responsibilities on the team include researching, developing, and deploying Machine Learning models and systems. Bazyli completed a M.Eng. degree in Engineering Systems and Computing at University of Guelph in 2016. During his degree he studied Data Mining, Wireless Sensor Networks, and Reinforcement Learning. Bazyli has experience with various Machine Learning techniques including Supervised, Unsupervised, Reinforcement, Deep, and Transfer Learning.”


Putting Machine Learning to Work for Digital Forensics

Digital devices and connectivity are creating a data tsunami for investigators in the law enforcement and corporate communities. Not only has the popularity of chat and email applications resulted in a massive spike of text-based conversations; but an increasing number of devices (smartphones, digital cameras, other IoT devices, etc.) also facilitate the easy capture of photos and videos. As a result, the number of pictures and messages that may appear in a case has grown exponentially, making it difficult for forensic examiners, investigators and analysts to keep up. The manual analysis of the evidence for context and relevance, combined with report building, can add significant time to an investigation. Magnet.AI allows an examiner or investigator to better prioritize their time in an investigation, and find potentially relevant evidence faster than they could through manual review.

In this talk we’ll introduce Magnet Forensics, a Waterloo-based software company, and how we’re helping to tackle this problem. As well, we’ll discuss how we’ve applied machine learning techniques to assist in the analysis.


Friday, March 15, 2019


2 p.m.


LH 3058 (Lazaridis Hall)

Unknown Spif - $key