My Story of Integrating Human Factors into Machine Learning for Driver State Monitoring
Over the past decade, driver monitoring systems have evolved from lab prototypes to automotive-grade products that save lives on the road. I was fortunate to contribute to this breakthrough, integrating human factors into machine learning for real-world impact.
Texas A&M University (2010-2016): Where My Journey Began
My journey started with my doctoral research on human factors and cognitive systems, where I explored novel metrics to enhance human performance in complex environments. Under Thomas Ferris’ supervision, I proposed “cognitive efficiency” - a metric quantifying how much information a person processes per “unit” of cognitive resources. A core chapter of my thesis involved measuring cognitive workload using physiological sensors. To deepen this research, I took graduate courses across multiple areas - computer science, statistics, psychology, and kinesiology - which fueled my curiosity about interdisciplinary connections. In my final year, a paper from Birsen Donmez’s lab, “Smart driver monitoring: When signal processing meets human factors”, opened my eyes to an exciting future: the intersection of human factors and machine learning to create intelligent human sensing.
UC Berkeley (2016-2017): A Brief but Pivotal Chapter
As a postdoctoral scholar, I collaborated with Steven Shladover, a pioneer of intelligent transportation systems, to study driver-automation interaction in truck platooning. It was an unforgettable experience of operating a three-truck platoon on Bay Area freeways. But Berkeley offered even more. During the kickoff meeting of the NSF VeHiCal project, I met Ruzena Bajcsy, a renowned Computer Science professor, who emphasized the importance of human factors for human sensing. Surrounded by several high-profile Berkeley professors, I was, however, too nervous to contribute - yet that moment profoundly shaped my vision of connecting human factors with machine learning.
By 2017, human factors was mostly framed as UI/UX in Silicon Valley, which didn’t fully align with my goal. Then, an unexpected opportunity arose on the other side of the world - Seeing Machines, an Australian company, was transitioning from eye tracking to driver monitoring. It needed human factors experts to help computer scientists understand driver state and behavior. After interviewing with Mike Lenné, their Human Factors Lead, and consulting Steve and Tom, I took a leap of faith - moving to Australia to embrace a new adventure.
Seeing Machines (2017-2024): Where Research Met Real-World Impact
Joining Seeing Machines in its ascent, I led the CANdrive Automated vehicle trial (AU$1.35 million) and later the collaboration with the MIT Advanced Vehicle Technology consortium (founded by Bryan Reimer). I initiated cutting-edge research on distraction, drowsiness, takeover readiness, and gaze-context analysis. This research demonstrated the unique value of human factors in defining ground truth, curating high-quality data, and enhancing feature engineering for driver state detection. Moreover, I explored computational models - from probabilistic machine learning to deep learning to generative AI - to analyze driver behavior and dynamic contexts. The findings offered novel and holistic insights into driver-automation-environment interactions, shaping the future of driver state monitoring.
Mutinex (2025-2026): The Same Methods in a New Domain
Every journey has its end. At the end of 2024, my time at Seeing Machines came to an end due to the volatility of the automotive industry. I knew this also meant the end of my career as an automotive human factors researcher - Australia doesn’t build cars.
However, it also meant a new start. I joined Mutinex as a Senior Data Scientist, building Bayesian machine learning models for marketing measurement. The domain was new, but the methods were familiar: the probabilistic modeling I had used to understand drivers now helped explain human decisions at commercial scale. It also showed me that what I had learned from studying drivers could travel well beyond automotive.
The Next Chapter
Today, my journey continues towards a broader future of empowering AI to support human tasks. I hope my story will inspire others who are passionate about human factors and AI for real-world influence.