DATA SCIENTIST & AI RESEARCHER
Data Scientist at Actual Reality Technologies·Ohio, USA
I build computer vision and machine learning systems that run on factory floors, from diffusion-based defect synthesis and segmentation to real-time process control.
I am a Data Scientist and AI Researcher at Actual Reality Technologies, where I lead computer vision and machine learning work for industrial clients: detecting defects that human inspectors miss, reading structural engineering drawings automatically, and optimizing furnace operations for energy and emissions.
My research focuses on the hard end of applied vision. Diffusion models (DDPMs) to synthesize the rare defect samples that imbalanced manufacturing datasets never contain. U-Net and vision transformer architectures for segmentation. YOLO-based detection pipelines that hold up outside a benchmark.
I take projects end to end: designing the pipeline, training and tuning the models, mentoring teammates through the full lifecycle, and working with leadership to turn a research result into something that survives contact with production. I hold an M.Sc. in Applied Statistics and a B.Sc. in Electrical Engineering, and I publish peer-reviewed research in computer vision and applied machine learning.
M.Sc. Applied Statistics (Business Analytics)
Bowling Green State University, Ohio
May 2023
B.Sc. Electrical Engineering
Azad University of Najafabad, Esfahan
Sep 2012
Four of these are live, running in your browser right now. Every visual is procedurally generated to illustrate the technique, not client data.
Generative Vision · Published Research
Manufacturing defect datasets are pathologically imbalanced: thousands of good samples, a handful of the failure mode you actually care about. I trained denoising diffusion probabilistic models to synthesize realistic rare defects, then used them to rebalance training and lift detection on the classes that matter.
Object Detection · Document AI
Engineering drawings are dense, inconsistent, and enormous. This pipeline combines CNN classifiers with YOLO detection to find and extract structural components across full drawing sets, turning a manual takeoff process into an automated one. I designed the full architecture independently.
Segmentation · Healthcare
Collaborative research across MRI brain tumor classification, lung cancer segmentation, and dental condition detection from panoramic X-rays. U-Net variants with CNN backbones and vision transformers, evaluated against clinical rather than benchmark criteria.
Time Series · Control
Machine learning optimization of furnace and melter operations: multivariate forecasting, operational persona clustering, and multi-objective control that improves energy and emissions without sacrificing product quality. Extended toward digital twin simulation and RL-based real-time control.
Predictive Analytics · Vision
Near-miss analysis and worker risk scoring from operational and environmental data, paired with vision-based detection of required safety equipment on site. Prediction feeding prevention, rather than post-hoc reporting.
LLMs · Research
Research quantifying label-induced bias when large language models evaluate themselves and each other, plus work on building genuine decision intelligence through iterative dashboard refinement rather than more charts.
7 papers across computer vision, medical imaging, and language model evaluation. 4 published, the rest under review.
A. Golkarieh, B. Afjehsoleymani, K. Kiashemshaki, S. Rezvani Boroujeni
BMC Oral Health (BioMed Central)
S. Rezvani Boroujeni, H. Abedi, T. Bush
Computer and Decision Making: An International Journal (COMDEM)
A. Golkarieh, S. Rezvani Boroujeni, K. Kiashemshaki, M. Deldadehasi, et al.
Computer and Decision Making: An International Journal
M. S. Hosseini, N. Raoofi, S. Rezvani Boroujeni, H. Najafzadeh
International Journal of Computational Intelligence Systems (Springer)
A. Golkarieh, K. Kiashemshaki, S. Rezvani Boroujeni, N. A. Isakan
Scientific Reports (under review)
M. Saraf, S. Rezvani Boroujeni, J. Beaudry, H. Abedi, T. Bush
arXiv preprint
L. Tadakala, M. Saraf, S. Rezvani Boroujeni, H. Abedi, T. Bush
arXiv preprint
AAA Club Alliance
Bowling Green State University
Araz Exir Trading Co.
The stack I reach for, and a few things people have handed me along the way.
05 / CONTACT
Open to conversations about applied computer vision, industrial AI, and research collaboration.
sajjadr@bgsu.edu