AMD Silo AI

the World Model Foundry is taking shape we're opening 4 PhD positions across UniBo and UniMoRE. research focus on world-action models, spatial AI, 3D scene understanding, and VLAs for humanoids and autonomy. for info, reach out to Matteo Poggi and Lorenzo Baraldi the World Model Foundry is a collaboration between AMD Silo AI Gbionics Alma Mater Studiorum – Università di Bologna Università degli Studi di Modena e Reggio Emilia AImageLab and many more to come apply here: https://lnkd.in/enPkUcgK (UniBo) and https://lnkd.in/dnkDVe5S (UniMoRE) Alessio Del Bue Daniele Pucci Marcella Cornia Sara Sarto

We're thrilled to welcome Giorgos Tolias, Associate Professor at CTU Prague and team lead of the Visual Recognition Group, to the next AMD AI Research Club session. Giorgos is no stranger to pushing the boundaries of computer vision, his work spans visual representation learning, vision-language models, and large-scale object retrieval. His recent ICCV paper was selected as a highlight paper, placing it in the top 3% of all submissions. He's also a 7× outstanding reviewer at CVPR and ICCV. In this session, hosted by @Pier Luigi Dovesi, we will discuss this highlight paper and some recent unpublished work, focusing on: 🕵️ Hidden metadata lurking inside visual embeddings  📷 Compression artifacts & camera pipeline signals  ⚠️ How shortcut learning quietly shapes AI predictions 📅 May 28, 2026 ⏰ 7 AM CT | 2 PM CET If you work in computer vision, representation learning, or AI robustness, this one is for you.  👉 Save your spot and come ready with questions. Link in the comments. #AIResearch #ComputerVision #FoundationModels #AMD #AIResearchClub

Advancing computational biology requires compute infrastructure that can keep pace with the scale and complexity of modern AI workloads in life sciences. AMD is a proud Gold Sponsor of #RECOMB2026, joining the global computational biology community to showcase how AMD compute is enabling companies and scientists to take full advantage of AI capabilities across genomics, structural biology, and medical imaging. Stop by our demo station to see AMD-accelerated AI in action, featuring: • OpenFold3: next-generation protein structure prediction • SwinUNETR: state-of-the-art medical image segmentation At AMD Silo AI, we work with leading life sciences companies to accelerate their AI workloads on AMD compute platforms. Come meet our experts at RECOMB in Thessaloniki, Greece, on May 26-29.  #TogetherWeAdvance_AI   #RECOMB2026   #LifeSciences  #HPC 

Getting started with AI applications shouldn't mean starting from scratch every time. AMD Solution Blueprints, built with AMD Inference Microservices (AIMs), are ready-to-deploy, customizable reference applications designed to accelerate the path from concept to deployment. Packaged as Helm charts and powered by AIMs and AMD compute platforms, they provide off-the-shelf microservice solutions for a range of use cases, from standard chat interfaces to multi-agent agentic frameworks, while giving teams the flexibility to tailor each reference application to their specific needs. Read more through the link below. #TogetherWeAdvance_AI  #LLMInference  #AIInference 

We are launching a research collaboration on Spatial AI with Alma Mater Studiorum – Università di Bologna and its Department of Computer Science and Engineering (DISI), home to CVLab, a leading European research group in stereo vision, depth estimation, and 3D reconstruction. The collaboration will focus on building geometry-aware perception modules on the AMD ROCm™ open software stack, bringing explicit 3D geometry into the Vision-Language-Action (VLA) and world-model pipeline for robotics and autonomous driving. Our vision is to turn frontier geometry research into open, high-performance software that helps advance the broader Physical AI ecosystem. This collaboration establishes CVLab as the Spatial AI node of a two-node academic hub alongside Università degli Studi di Modena e Reggio Emilia, feeding into the World Models Foundry, the AMD Silo AI open Physical AI research program. Learn more about the collaboration through the link in the comments. #TogetherWeAdvance_AI   #WorldModels   #PhysicalAI 

What hidden signals does your camera leave behind in AI embeddings? Join the next AMD AI Research Club session as Pier Luigi Dovesi hosts Giorgos Tolias, Associate Professor at CTU Prague, for a deep dive into acquisition traces in modern visual foundation models. 📅 May 28, 2026 ⏰ 7 AM CT | 2 PM CET We’ll explore: • Hidden metadata inside visual embeddings • Compression artifacts and camera pipeline signals • How shortcut learning impacts AI predictions Register through the link in the comments!

AI inference can be computationally intensive with resource requirements that vary depending on traffic, e.g., the number of inference requests your workload receives at any given time. #Autoscaling of LLM inference allows for optimized resource utilization to manage fluctuating AI Workload demands in AMD Inference Microservice (AIM) deployments. Through the AIM Engine and AMD AI Workbench, autoscaling addresses the challenge of maintaining latency or throughput of the inference service when demand fluctuates. Resources automatically scale up during peak traffic to maintain performance and scale down during quieter periods to reduce cost and resource consumption. Read more through the link below. #TogetherWeAdvance_AI  #LLMInference  #AIInference 

#SUSECON2026 is in full swing with great conversations happening around enterprise AI, sovereign infrastructure, and open-source innovation! With keynotes and panels by SVP Keith Strier and Corporate VP Kumaran Siva, and breakout sessions and demos by our engineering team Isabelle Liu, Andy Allred and Harry Souris, we are showcasing how to put open source to work at scale while maintaining strategic control over your AI capabilities. We’re excited to connect with so many teams building the future of sovereign enterprise AI, drop by the AMD booth to say hello! 👋 #TogetherWeAdvance_AI  #OpenSource  #SUSE 

AI sovereignty requires strategic control across data, models, supply chain, and infrastructure. At #SUSECON26 AMD Corporate VP Kumaran Siva will keynote on how enterprises and governments can maintain full ownership over their AI capabilities while ensuring security, compliance, and resilience. By increasing supply chain resilience though open ecosystems, organizations can control their AI future while driving economic growth and protecting critical data. Learn more at SUSECON by SUSE, April 20-23. #TogetherWeAdvance_AI  #OpenSource  #SovereignAI