
Founder & CEO |NCA
Applied AI researcher publishing on LLM inference optimization on AMD GPUs and multimodal document retrieval. Builds production RAG and agent systems.
GitHub · 26 repos · 252 stars
Hugging Face · 7 models · 125.5k downloads
Biography
Athos Georgiou is the Founder and CEO of NCA, a company focused on AI research and consulting. Born in Cyprus, Athos began his career as a Technical Mechanical Engineer at the Cyprus University of Technology, where he laid the groundwork for his future endeavors in AI. His journey in AI began with a deep frustration with the repetitive coding tasks he encountered, which eventually led to the founding of SUPPER London, a luxury food and drink concierge service. This experience in logistics and customer service provided a unique perspective that he later applied to his AI ventures. In 2026, Athos co-authored the paper "Hydra: Unifying Document Retrieval and Generation in a Single Vision-Language Model," which was published on arXiv. This work showcases his expertise in developing AI models that can handle both document retrieval and generation tasks, a significant contribution to the field of AI. Additionally, Athos has been involved in the development of the Hydra project, which aims to unify these tasks in a single vision-language model, demonstrating his commitment to advancing AI technology. Currently, Athos leads NCA, where he focuses on AI research and consulting. His work at NCA includes projects such as "Athos-Graphene," a computational discovery of a 2D carbon anode for lithium-ion batteries, which highlights his interdisciplinary approach to AI. Athos is also a co-author of the paper "Architecture-Aware LLM Inference Optimization on AMD Instinct GPUs," which was published in 2026. This paper discusses the optimization of large language models (LLMs) for GPU inference, a critical aspect of AI performance engineering. What makes Athos unique in the AI/tech landscape is his ability to bridge the gap between technical innovation and practical application. His work on projects like Hydra and Athos-Graphene exemplifies his dedication to creating impactful AI solutions that address real-world challenges. Athos's unique perspective, gained from his diverse career
Hydra is a groundbreaking vision-language model that unifies document retrieval and generation tasks. This model has significantly advanced the field by integrating both retrieval and generation capabilities, making it a valuable tool for businesses and researchers.
This paper introduces a novel method for spatially-grounded document retrieval, enhancing the precision and relevance of document search results. The technique uses patch-to-region relevance propagation to improve the accuracy of visual document retrieval tasks.
This study provides a comprehensive analysis of optimizing large language model inference on AMD Instinct GPUs. It offers valuable insights and benchmarks for deploying and optimizing LLMs on specific hardware, which is crucial for both research and industry applications.