Omer Tariq

I am an AI Research Scientist specializing in 3D/4D perception, vision-language grounding, and model efficiency for embodied AI. Over the past decade, I have built deep expertise spanning from hardware-level FPGA design to state-of-the-art foundation models, recently completing my Ph.D. in Computer Science at KAIST in 2025. My work focuses on making massive multimodal understanding—such as spatial reasoning and open-vocabulary recognition—viable under real-time, power-constrained edge deployment. I have a strong track record of transitioning research into deployed systems. Two key achievements include my work at Neubility, where I developed a production multi-camera pipeline that unifies 3D object detection, monocular depth estimation, and semantic segmentation behind a single DINOv3 backbone. This system achieves 25 FPS across a 5-camera array on a single Jetson Orin SoC. Additionally, I extensively worked in Neural SLAM, visual-inertial SLAM, navigation and path planning for autonomous mobile robotics applications. I am seeking a senior or lead remote role in applied AI research, machine learning engineering, or autonomous systems. I thrive in environments that challenge me to bridge the gap between heavy multimodal architectures and efficient, real-world edge deployment, and I am eager to bring my expertise in quantization, knowledge distillation, and spatial reasoning to a forward-thinking remote team.