Mona Jalal
Senior Computer Vision Research Engineer at Toyota Material Handling North America
About
I am Mona Jalal, a Senior Computer Vision Research Engineer at Toyota Material Handling North America. My career has been a journey through the intersection of deep academic research and real-world industrial application, having moved from PhD research at Boston University and internships at NVIDIA and Twitter into the Advanced R&D space. I specialize in 3D vision, 6D pose estimation, and bridging the 'Sim-to-Real' gap to make robotics more reliable in complex environments like warehouses. I am deeply passionate about moving spatial computing forward and advocating for normalcy-based learning in defect detection. Beyond the lab, I am a strong supporter of the growing tech ecosystem in Indiana. I love connecting with fellow researchers and engineers to discuss how we can reduce the friction between perception and actuation to solve the next generation of automation challenges.
Networking
What I can offer
- ›Technical insights into deploying AI in complex industrial environments
- ›Expertise in 3D scene reconstruction and robotics perception
- ›Guidance on Sim2Real workflows and synthetic data generation
Looking for
- ›expanding my professional network
- ›exploring mutual opportunities in robotics and computer vision
Best fit for
Current Interests
Background
Career
Transitioned from academic research at Boston University and UW-Madison through internships at NVIDIA, Twitter, and DawnLight into senior industrial R&D at Toyota.
Education
MS Computer Science (Boston University), MS Computer Science (UW-Madison), MS Electrical Engineering (UW-Madison), MS Computer Engineering (Sharif University of Technology), BS Computer Engineering (National University of Iran).
Achievements
- ›Recipient of O-1A Visa (Extraordinary Ability) and NIW U.S. Permanent Residency
- ›Inventor with multiple patents in industrial AI and perception systems
- ›Area Chair and Workshop Organizer for CVPR, ICCV, WACV, and NeurIPS
Opinions
- Industry relies too heavily on rare defect samples; advocates for normalcy-based learning like PatchCore.
- Spatial computing must move beyond '3D UI in a box' toward meaningful interaction primitives.
- Object-centric reconstruction is superior to monolithic scene guesses for production reliability.