Mahdiyar Molahasani

Architect of Invisible Intelligence

I don’t build models that impress for a season. I forge AI that conquers industries and then vanishes into their bloodstream silent, unbreakable, inevitable

ML Researcher II at Captura PhD, Queen's University Computer Vision + Generalization Vancouver, BC

ML Researcher II at Captura, where I build computer-vision and multimodal systems for the photography industry. PhD from Queen's University; published at ICCV, AAAI, and NeurIPS Workshop. I translate raw intelligence into production empires that survive domain shifts, regulatory storms, and human chaos.

Vision-Language Models Federated Learning Domain Generalization Fairness Continual Learning

View Research Deep Dives Get in Touch

Experience Timeline

Industry + research trajectory

Jun 2025 - Present

Current

ML Researcher II

Captura · Vancouver, BC, Canada (Based in Raleigh, NC, USA)

Leading applied ML research and development for AI image enhancement and management for the volume photography industry.

Build reusable ML infrastructure and evaluation tools for computer vision experimentation.
Prototype product-facing ML features and improve research-to-deployment workflows.
Shape data strategy, labeling workflows, and cross-functional ML capability.

May 2024 - May 2025

Remote AI Evaluation

Expert LLM Trainer & Evaluator

Outlier AI · Remote from Kingston, ON, Canada

Evaluated multimodal AI outputs and designed advanced validation tasks for model accuracy, reliability, and reasoning quality.

May 2021 - May 2025

Doctoral Research

Graduate Research Assistant

Queen's University · Kingston, ON, Canada

Completed PhD research in computer vision, federated learning, fairness, continual learning, and domain generalization, publishing in venues including ICCV, AAAI, ICASSP, ICIP, and NeurIPS workshops.

May 2019 - Apr 2021

ML Foundations

Graduate Research Assistant

University of Saskatchewan · Saskatoon, SK, Canada

Developed deep learning methods for medical imaging, super-resolution, COVID-19 detection from chest X-rays, and dental caries detection.

A Working Belief

What I’m Architecting

The next generation of meaningful AI companies will not be built by chasing models alone. They will come from people who can translate intelligence into products, workflows, and institutions that change how industries operate.

The best AI systems do not feel impressive for long. They become invisible. They remove friction, absorb complexity, and quietly make the work better. That gap, between a model and a system that quietly works, is where I spend my time.

Research Impact

609 Citations

18 Papers

10 H-Index

AI scientist with 8+ years of research experience in vision-language models, federated learning, and domain generalization. I bridge the gap between academic innovation and production-grade computer vision systems for the photography industry.

Applied Research & Systems

Bridging the gap between theory and production

Semantic Search & VLM Infrastructure

Architected CLIP-powered semantic sorting and multimodal understanding pipelines, powering large-scale image management for professional photography collections.