Payam Mousavi
Physicist / Applied ML Scientist
I am a multi-disciplinary scientist/engineer who thrives on exploring connections between different domains and industries. I have an academic background in physics and engineering with many years of experience developing software and ML-based solutions for multiple industries such as supply chain, manufacturing, oil & gas, advertising, biology, defense & security, and most recently AI/ML consulting
- Location
- Vancouver, British Columbia, Canada
- Website
- https://payam-mousavi.com
- PayamMousavi4
- pmousavi
Experience
– present
Applied Research Scientist at Alberta Machine Intelligence Institute (AMII)
Developing and deploying ML-based solutions for various industries such as supply chain, manufacturing, oil & gas, advertising, biology, defense & security as well as conducting applied research
Highlights
- Leading the advanced technology group to develop software solutions for industrial clients as well as other departments within Amii
- Researching Physics-Informed Neural Networks (PINNs) and their applications to fluid flow
- Applying RL to industrial control, multi-robot planning for logistics applications, and VLSI routing
- Developing hybrid Operations Research and RL algorithms for the optimization of a 3D warehouse structure with multiple interacting robots
- Developing machine vision models for detection and classification of gas emissions
–
Senior Data Scientist (R&D Director) at Unbounce Marketing Solutions
Leading the R&D team to develop tools, leveraging deep learning, statistics, classical ML, causal models, and RL to move forward the company strategy in “Conversion Intelligence” within a digital marketing eco-system
Highlights
- Developing machine vision models for marketing applications
- Developing NLP models to generate/classify text that maximize the conversion rate
- Exploring Reinforcement Learning for designing high-conversion web/landing pages
–
Staff R&D Scientist at MDA Systems
Applying deep learning to to natural images, Earth Observation (EO), and Command & Control (C2)
Highlights
- Designed/implemented GANs and VAEs (Python/PyTorch/TensorFlow) to synthesize and manipulate imagery and to perform anomaly detection
- Implemented, Supervised (ResNet-based), and Semi-Supervised (FixMatch) models (in PyTorch) for image classification and detection (RetinaNet and Faster-RCNN) of vessels and planes in satellite imagery
- Applied Multi-agent RL in a cooperative setting for applications in Defense (i.e., Command & Control) and surveillance.
–
Research Scientist at Phase Technology
Building optical analyzers for measurement of cold flow properties (mainly of oil and gas)
Highlights
- Designed/optimized optical imaging systems (TracePro, COMSOL, OpenCV, and MATLAB)
- Developed software (MATLAB and Python) for robotic arm manipulation for sample loading
- Used various machine learning techniques for sample classification
–
R&D Scientist (PhD Candidate) at Honeywell Process Solutions (ACS)
Building optical analyzers for measurement of cold flow properties (mainly of oil and gas)
Highlights
- Designed/optimized optical imaging systems (TracePro, COMSOL, OpenCV, and MATLAB)
- Developed software (MATLAB and Python) for robotic arm manipulation for sample loading
- Used various machine learning techniques for sample classification
Education
–
PhD in Physics from Simon Fraser University
–
MSc in Mechanical Engineering from University of British Columbia
–
BASc in Engineering Physics from University of British Columbia
Publications
Human-in-the-Loop Reinforcement Learning: A Survey and Position on Requirements, Challenges, and Opportunities , Journal of Artificial Intelligence Research (JAIR)
RL-Ripper: A Framework for Global Routing using Reinforcement Learning and Smart Net Ripping Techniques , Proceedings of the Great Lakes Symposium on VLSI (GLSVLSI)
MaskRenderer: 3D-Infused Multi-Mask Face Re-enactment , arXiv: 2309.05085
Maximum Likelihood parameter estimation in terahertz time-domain spectroscopy , Optics Express
A Real-time Bayesian Decision-Support System for Predicting Suspect Vehicle’s Intended Target Using a Sparse Camera Network , International Conference on Defense , Security, Intelligence (ICDSI)
Deep Learning for Vessel Detection and Identification from Spaceborne Optical Imagery , ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences
Onboard Artificial Intelligence for Space Situational Awareness with Low-Power GPUs , 21st Advanced Maui Optical and Space Surveillance Technologies Conference
Human-AI Teaming with the Digital Battlespace Framework , 25th ICCRTS International Command and Control Research and Technology Symposium
Simultaneous composition and thickness measurement of paper using terahertz time-domain spectroscopy , Applied Optics
Chipping into microfuidics , Physics World
A novel flow reactor for studying reactions on liquid surfaces coated by organic monolayers: Methods, validation, and initial results , The Journal of Physical Chemistry A
Continuous referencing for increasing measurement precision in time-domain spectroscopy , US Patent 8378304 B2
Time domain spectroscopy (TDS) based method and system for obtaining coincident sheet material parameters , US Patent 8187424
Languages
- English
- Fluency: Native speaker
- Farsi
- Fluency: Native speaker
Skills
- Machine Learning
- Level: MasterKeywords:
- Mathematical Modeling
- Level: MasterKeywords:
- Physics
- Level: MasterKeywords: