Khashayar Ghamati

I’m Khashayar -- a Principal AI Engineer at Inframatic Engineering, where I design and deploy adaptive AI systems for automated compliance in UK. My focus is on learning agents that operate and adapt in real time within dynamic, non-stationary environments -- bridging advanced research in online learning, reinforcement learning, continual learning, and meta-learning with scalable, production-ready systems.

Alongside industry, I’m in the final year of a PhD in Computer Science at the University of Hertfordshire. My research centres on NOVA, a cognitive framework for continual learning, novelty detection, and reinforcement learning in human–robot teaming. The work has been validated on the ARI humanoid platform at the UH Robot House and spans bio-inspired gaze control (RLBAM), confidence-aware perception (CAPA-AI), and activity recognition (HAR-Agent), with publications in venues including IEEE Transactions on Cognitive and Developmental Systems.

I also supervise BSc and MSc students at the University of Hertfordshire on AI and robotics projects -- a role I enjoy because it keeps me close to the first-principles questions that get lost when you spend too long shipping. The rhythm I care about most is the one between research and engineering: translating ideas into systems that hold up under production load, and letting real-world constraints shape the questions worth pursuing in the lab.

My technical interests sit across continual learning, agent architectures, MCP and multi-agent systems, MLOps, and the kind of MLE work that turns a paper into something reliable. I’ve been writing Python for over a decade, and I’m most at home at the seam where theory meets what actually ships.

If you’d like to get in touch about research collaboration, speaking, or anything at the intersection of adaptive AI and real-world systems -- say hello.