I'm a PhD researcher at RPTU and DFKI working on trustworthy AI and generative models for multimodal document understanding. More recently, my interests have expanded toward LLMs and agentic systems, and toward building more general and open-ended AI systems. My research has resulted in 15+ publications, and I open-source much of the software I build, including TorchXAI and Atria.

Before my PhD, most of my work was in computer vision and robotics, building real-time vision systems and working with autonomous mobile robots, UAVs, and humanoid robots. My master's research focused on developing a generalized whole-body motion framework for humanoid robots. I also led NUST's RoboCup SPL team during this time, where we worked on getting humanoid robots to play soccer autonomously, covering perception and localization, motion and control, and coordination between robots.

My interests in robotics and modern AI come from the same underlying curiosity about machine intelligence and consciousness, and about how intelligent robots might eventually become part of everyday human life. With recent progress in LLMs, world models, and embodied AI, some of the problems that originally drew me to robotics are starting to feel much less distant. It's a particularly interesting time to be working at the intersection of these areas.

Outside of work, I'm usually gaming, watching anime, or hanging out with friends on Discord.