My name is Mariya, and I'm an AI Research Scientist working on agentic safety, multi-agent collaboration and communication, long-horizon reasoning, and behavioral alignment in emergent model capabilities. I'm very interested in understanding how intelligent systems solve complex problems under partial observability, share information and coordinate across specialized capabilities, and adapt their behavior to human preferences and intentions.

During my time as a Senior Research Scientist at Meta Superintelligence Labs (MSL), I developed evaluation and scalable oversight methodologies for frontier multimodal AI systems. My work focused on characterizing safety and alignment failures across vision-language models, translating safety policies into measurable evaluation criteria, and building large-scale benchmarks and automated grading systems calibrated against human judgments.

Prior to that, I was an Applied Scientist at AWS Responsible AI, where I developed evaluation methodologies for auditing bias, privacy, and robustness in computer vision and multimodal foundation models, with a focus on application-specific fairness metrics, membership inference defenses, and evaluation of model guardrails in large-scale deployments. Over the years, I've worked on vision-language models, multimodal representation learning, image and video generation, agentic reasoning and evaluation, interpretability, and synthetic data generation from minimal human supervision.

I have spent several years in industry research working on a broad range of vision and multimodal areas, including image and video generative models, visual recommender engines, 2D-to-3D human body shape and pose modelling, synthetic data generation for efficient training of foundation models at scale.

I received my PhD in Computer Science from the University of Illinois at Urbana-Champaign, where I was advised by David A. Forsyth. My doctoral research focused on problems in vision and language, visual search and retrieval, multimodal generation, and applications of computer vision in the fashion domain.

Outside of research, I spend an inordinate amount of time lifting heavy weights, as well as filming, editing, and color-grading candid scenes from my travels and film walks. I also maintain an objectively impressive collection of vintage clothing, sourced from thrift stores around the world — a habit that pairs suspiciously well with conference travel.

If you're interested in collaborating or just want to connect, feel free to reach out.