Summary
Mojtaba Marvasti is a Computer Vision Scientist with a PhD and nearly a decade of focused experience building and deploying advanced ML/CV systems across remote sensing, aerial imagery, and multimodal perception. He has authored 25+ papers including highly cited survey articles in IEEE T-ITS and ACM Computing Surveys, and led impactful research like a Forest Health Monitoring project in collaboration with Natural Resources Canada. Technically fluent in PyTorch, TensorFlow, OpenCV and MLOps toolchains (Azure ML, MLflow, SageMaker), he designs efficient CNN/Transformer-based models, PEFT strategies, and lightweight architectures for real-time, multi-sensor deployment. His work spans cutting-edge areas—multimodal LLMs, vision foundation models, diffusion/GANs, semi-supervised learning, NAS, and explainable AI—bridging theoretical insight with production-ready solutions. A proven mentor and cross-disciplinary collaborator, he has guided students to top theses and competitive challenge placements (VisDrone top rankings). Less obvious: he combines deep academic rigor with hands-on edge optimization and data-standardization expertise for thermal, SAR, and drone-collected datasets.
9 years of coding experience
6 years of employment as a software developer
Ph.D. Computer Vision, Ph.D. Computer Vision at Yazd University
English, Persian