From visions of film to computer vision. Jason morphed his passion for filmmaking, where actors attempt to convince the viewers of genuine emotion, into a career of programming computers to interpret the believability of genuine facial expressions.
Jason started his journey in Charlotte, North Carolina. He followed his passion and transferred to the University of North Carolina Wilmington in 2007 to enroll in film studies. Working as a production assistant in film, Jason recognized a strong interest in computer science and made a final transition into a CS degree. With a Ph.D. in the back of his mind, he found his way into a biometrics lab as a sophomore and grew curious about facial recognition. Jason decided to pursue and complete his master's, actively involved in many projects, publishing his work as a sophomore and junior. His research focused on understanding the difference between a posed smile and a genuine smile inspired by an interactive virtual human he stumbled upon once at a conference.
With all of his research and publications under his belt, he went on the hunt for a university where he could complete his Ph.D. in 2012. He quickly fell in love with Cardiff University in Wales. At the time, most research was focused on basic conversational expressions like anger and happiness. Jason wanted to focus his research on more complex conversational expressions like confusion, thinking, agreement, and pleasant surprise. He started working with the school of psychology and computer science to create statistical models to predict what the listener in a conversation would do in response to the speaker. As he finished his Ph.D., he discovered the work that Dr. Mark Sagar was doing at the University of Auckland on hyper-realistic digital humans. Jason moved to New Zealand to build simulated brains and nervous systems that would not just mimic humans but respond organically. They read books to digital children to teach them about animals, observed how mothers and children interacted and learned over time, and provided functionality to allow people with disabilities to access information about their benefits. Jason utilized deep learning models to build prototypes of these systems. This research spun off into a startup called Soul Machines which helps brands utilize the interaction of human and machine collaboration.
Jason then made the transition back to the US and from academia into industry where he started as a Research Scientist at Siemens. He worked on designing systems to apply deep learning techniques for automating engineering environments; and understanding human behavior in industrial environments, to improve safety, efficiency, and training outcomes. He authored three patents in these areas while at Siemens. He then joined Duke Energy as a Deep Learning Engineer where he lead a team to create a multi-camera, real-time object identification system for analytics, which was used in interactive kiosks in smart cities around the US. Not only did he design and build the prototype from scratch, but he also assembled a team that helped grow the project to be one of the most important internal technical projects at Duke Energy.
Jumping from energy to professional motorsports, Jason joined Rho AI as their Senior Data Scientist. Not only did he focus on increasing the speed of building and deploying data science solutions, but he supported the team that won the 2020 NASCAR Championship. Talk about fast. The solutions he worked on enabled the pit crew chief to decide the optimal time that his team should stop and make adjustments, like adding tires or fuel, based on the likelihood of actions that other teams might make. Dashboards took on a whole new meaning. Jason then joined Lapetus Solutions to design the architecture for their MLOps pipeline to address the challenges that the machine learning developers were facing when deploying and managing their models. He developed a modular, automated, and scalable solution for the developers to interface with. He then went on to work for an accelerator that provided data science services to startup companies.
Now, Ikonos is lucky to have Jason on the team as our Senior MLOps Architect and Instructor. He uses his industry experience and academic knowledge to help our customers improve and scale their MLOps capabilities. Jason has a strong background in applying machine learning techniques to an extremely diverse range of problem domains. Of course, work is not the only thing Jason is passionate about. In his free time, you will find him using his 3D printer to build an RC car, building robots, traveling the world with his wife (and soon with his daughter), or coaching a soccer team.