Intel’s 2025 Outstanding Researcher Awards Honor 10 Academic Innovators

Intel’s 2025 Outstanding Researcher Awards (ORAs) recognize academic leaders whose breakthroughs are shaping what’s next, from foundational science to real-world impact.

The annual program spotlights exceptional work from Intel-sponsored university research, advancing Intel’s mission to create world-changing technology that enriches the lives of every person on Earth. Led by Intel’s Corporate Research Council, the ORAs drive collaboration with top universities, research centers, and ecosystem partners.

The winners of Intel’s 2025 Outstanding Researcher Awards are:

John Fowler / Arizona State University

AI Assisted Digital Twins for Supply Chain 

Creating and implementing supply chain digital twins continues to be expensive and time-consuming. This research utilized generative AI and reinforcement learning methods to automate simulation development, validation, and enhancement processes. An innovative two-phase framework featuring structured intermediate representations (employing a specialized LLM) substantially enhanced efficiency and precision compared to general-purpose LLMs, minimizing development resources and facilitating scalable, reliable digital twins for practical supply chain management.

Farzan Gity / Tyndall National Institute 

Role of Grain Boundaries (GBs) in TMDs on carrier transport: identifying GB structure-electrical property correlation 

This project established detailed connections between grain boundary configuration and electrical behavior in 2D transition metal dichalcogenide devices. By integrating modeling techniques with experimental characterization, it determined which grain boundary structures severely compromise carrier conduction in MoS₂ and WSe₂. The outcomes provide vital guidance for material engineering and device architecture, directly enabling the creation of enhanced 2D channel materials.

Cristiano Giuffrida / VU Amsterdam

Allocamelus: Efficient Memory Safety 

Delivering complete memory safety with minimal performance impact continues to be a persistent challenge in contemporary software systems. This research provided rapid and modular defense against spatial and temporal memory vulnerabilities by integrating secure memory allocation with runtime monitoring enhanced by current Intel processor capabilities. By utilizing features like Linear Address Masking and the floating-point unit, the solution scaled effectively to large, unchanged applications while identifying potential future hardware acceleration possibilities.

Jiayi Huang / Hong Kong University of Science and Technology

Software and Hardware Prepush Multicast 

As processor core counts increase and workloads become more data-intensive, on-chip bandwidth and latency face mounting challenges, especially when handling frequently accessed shared data. This research introduced an efficient multicast solution for last-level caches and on-chip networks using precise sharer prediction and an innovative pre-push mechanism. The approach proactively distributes shared data through a streamlined software-hardware interface, eliminating unnecessary traffic and substantially enhancing network-on-chip efficiency and application performance for AI and datacenter applications.

Mohsen Imani / University of California, Irvine

Efficient and Interpretable Symbolic AI for Intelligent Reasoning and Decision Making 

Next-generation brain-inspired computing approaches provide an attractive alternative to standard deep learning frameworks. This research enhanced efficient and transparent symbolic AI using hyperdimensional computing, demonstrating orders-of-magnitude improvements in training and inference efficiency, while preserving excellent reasoning capabilities. The work created new adaptable, energy-conscious algorithms at the convergence of hyperdimensional computing, probabilistic machine learning, reinforcement learning, and active learning, showing outstanding results with thousands-fold performance and power enhancements.

Vijay Raghunathan / Purdue University

Multimodal System Understanding and Heterogenous AI compute across CPU, GPU and NPU 

Next-generation AI systems increasingly require the coordinated use of diverse computing engines to deliver high performance and energy efficiency at scale. The research advanced heterogeneous AI computing across CPUs, GPUs, and NPUs for transformers, vision-language and vision-language-action (VLA) models, graph neural networks (GNNs), and newer AI architectures such as Mamba and state space networks (SSNs), including applications in multimodal and embodied AI. The work introduced new techniques for cross-accelerator coordination that improved performance and energy efficiency, while informing the evolution of Intel’s next-generation AI platforms and system architectures.

Visvesh Sathe / Georgia Institute of Technology

Runtime Thermal Management in Scaled 3D SoC Technologies – Circuits, Techniques and Architectures 

Effective thermal management in advanced 3D system-on-chip designs demands precise, low-overhead sensing solutions. This research developed compact digital temperature sensors powered by Vdd, featuring high precision, minimal noise interference, and simplified single-point calibration. Through validation across multiple test chips and firmware implementations, these sensors provide process-scalable, detailed thermal monitoring that ensures reliable and energy-efficient performance in cutting-edge semiconductor technologies.

Uwe Schroeder / NamLab gGmbH & Eilam Yalon / Technion – Israel Institute of Technology

Understanding and controlling the growth and properties of ferroelectric HZO-based device stacks 

Enhancing hafnia-based ferroelectric devices requires more profound understanding of fabrication processes, defect properties, and switching phenomena at the atomic dimension. This research center established complete insight into strain mechanics, defect formation, and electrode templating for preserving ferroelectric phases and directing polarization switching in GHz operating conditions. These breakthroughs boosted reliability by allowing optimized ferroelectric device configurations for varied high-frequency applications.

Shenlong Wang / University of Illinois at Urbana-Champaign

Harnessing Generative AI for Realistic, Interactive, and Physics-Grounded Virtual Environments

Constructing authentic, interactive digital twins at metropolitan scale necessitates innovations in both generative modeling and physical simulation approaches. This research developed computational methods and software architectures for comprehensive, photorealistic, and physics-informed virtual environments by merging generative models with multimodal mapping and physics-based simulation techniques. The developed platforms supported immersive digital twin applications and furthered Intel's strategic roadmap for 3D content creation and simulation computing.