About
Today
I’m CTO at MultiplAI, where I build AI-assisted compliance systems for regulated financial institutions.
My work focuses on document understanding, model validation, and trustworthy AI that operates with clear evidence, traceability, and human oversight.
What I work on
My work has moved from fundamental machine-learning research, through applied ML and quantitative finance, toward building AI systems that have to work reliably inside real organizations.
Across those experiences, I've become interested in the gap between doing the work and having enough to make a useful decision. An experiment can produce a valid result without answering the product question. A metric can be correct while missing something important. A detailed specification can still leave the actual decision unclear.
I write about these situations, particularly how to recognize what is missing, what is unnecessary, and what is sufficient to move forward.
Previously
Before MultiplAI, I worked at Quantfury on automated trading strategies using proprietary retail-activity data. I also modelled and deployed a reinforcement-learning agent for trading in live markets.
Earlier, at RBC Borealis, I worked across fundamental and applied AI research, including reinforcement learning, multi-agent systems, robustness, and explainability for trading systems.
Research
My academic work includes multi-agent deep reinforcement learning, auxiliary tasks for deep RL, robustness in reinforcement learning, and explainability for RL trading agents.