At Superside, I lead AI pipelines for image and video product photography that replace traditional photoshoots for brands including GAP, Dick’s Sporting Goods, and EssilorLuxottica.
I design brand-aware workflows that combine diffusion-based visual generation APIs, personalization, LLM prompting, and QA. I also drove the GenAI automation strategy for an EssilorLuxottica engagement worth approximately $1.5M.
Previously at Picsart, I led R&D for controllable diffusion-based image and video generation, taking models through training, fine-tuning, evaluation, and deployment. My research has appeared at CVPR, ICCV, and WACV, alongside open-source HuggingFace releases.
I also teach Artificial Intelligence, Decision Support, and Generative AI at the American University of Armenia. I bring a research-grounded approach to building reproducible, production-ready ML systems.
