Intel's 2021 Outstanding Researcher Awards Recognize 17 Academic Researchers


  • Intel's 2021 Outstanding Researcher Awards recognize 17 leading worldwide academic researchers conducting Intel university-sponsored research.

  • These research contributions advance today's computing by building upon Intel products and developing innovative technologies.


Intel is pleased to present its 2021 Outstanding Researcher Awards (ORAs) to 17 leading academic researchers. The annual award program recognizes the exceptional contributions made through Intel university-sponsored research. Intel sponsors and works alongside academic researchers around the globe in areas such as field-programmable gate arrays (FPGAs), artificial intelligence (AI), and other innovative technologies. 

Intel's ORAs are part of Intel's Corporate Research Council, which participates in research initiatives with prominent university science and technology centers, the National Science Foundation, ecosystem partners, and the Semiconductor Research Corporation. With projects such as bridging Intel® oneAPI to managed programming languages, to hardware/software contracts for more secure speculation, to patterning at nano-length scales by directed assembly, these Intel and university-sponsored research collaborations are advancing today's computing into future technologies. 

"Intel’s academic research partnerships are a core part of the company’s strategy to explore critical research paths," said Aravind Dasu, co-director of Intel's Corporate Research Council. 

"We are pleased to recognize the important contributions of these carefully selected researchers in our 2021 Outstanding Researcher Awards. We wish them each sincere congratulations," said Henning Braunisch, co-director of Intel's Corporate Research Council.

Left to right, top row: Christos Kotselidis, Marco Guarnieri, Onur Mutlu, Umit Ogras, Eric Pop

Middle row: Placid Ferreira, Shiv G. Kapoor, Tajana Rosing, Tim Kraska, Marian Verhelst, Wannes Meert

Bottom row: Guy Van den Broeck, Chinedum Osuji, Justine Sherry, Vyas Sekar, James C. Hoe, Ron Dror

The 2021 Intel Outstanding Research Award winners are:

Professor Christos Kotselidis, University of Manchester, UK
Bridging Intel oneAPI to Managed Programming Languages

This research bridges Intel oneAPI and managed languages. XPU features are exposed to the Java* platform through interfaces in SPIR-V*, Level Zero*, and the University of Manchester's TornadoVM*. The runtime dynamically strives to optimize code based on the target hardware, enabling dynamic reconfiguration of the application for an improved hardware/code combination.

Professor Marco Guarnieri, Madrid Institute for Advanced Studies in Software Development Technologies (IMDEA), Spain
Hardware/Software Contracts for Secure Speculation

This project formalizes hardware and software contracts to enable principled hardware/software co-design for more secure speculation. This, in turn, allows the software to reason about the specific types of defenses it needs to implement based on the hardware contract. This project won a Best Paper Award at the 42nd IEEE Symposium on Security and Privacy.

Professor Onur Mutlu, Swiss Federal Institute of Technology (ETH) Zürich, Switzerland
Efficient Compute-in-Memory Architectures

Professor Onur Mutlu’s emerging-workload-driven research has led to open-source benchmark suites that help hardware-software co-design and exploration for near-memory architecture. His team’s research has also led to innovative near-memory architectures for deep learning and genomics applications.

Professor Umit Ogras, University of Wisconsin-Madison, USA
Fast System-Level Performance Modeling for Emerging Applications

Designing best-in-class software and hardware for modern applications requires system-level pre-silicon simulation. However, detailed simulators are slow and cannot provide realistic results for modern and emerging workloads. An on-die communication fabric – a shared resource used by all IP blocks – is a central piece of the problem. In close collaboration with Intel, the team delivered automated modeling technology based on extending and applying queueing theory that delivers meaningful simulation speedup while maintaining acceptable accuracy with respect to detailed simulators for cache coherent server and client fabrics.

Professor Eric Pop, Stanford University, USA
Doping and Contacts for Atomically Thin 2D Semiconductors

Professor Eric Pop’s team focused on improving Moore’s Law scaling with two-dimensional (2D) materials by solving some of the important challenges in this field. The team made two discoveries that can influence future research in the field of 2D transistors. One is a predictive optical metrology which correlates with device performance. Another is a prototype of NMOS contact which is on par with silicon technology.

Professors Placid Ferreira and Shiv G. Kapoor, University of Illinois Urbana-Champaign, USA
Experimentation Management for Complex Multi-Step Manufacturing Process Flows

The UIUC team’s research goal is developing a methodology to perform systematic and coordinated end-to-end design-of-experiments life-cycle management for complex multi-step semiconductor processes. This includes the development of a theoretical approach to effectively identify subsets of unmonitored process variables responsible for shifts in monitored process output parameters.  

Professor Tajana Rosing, University of California San Diego, USA
MLWiNS: Hyper-Dimensional Computing for Scalable Intelligence Beyond the Edge

This project aims to leverage the benefits of hyper-dimensional computing (HDC) in a distributed Internet of Things environment. The inherent robustness of hyper-dimensional representations helps protect against the unreliability of wireless communication channels. The fast, energy-efficient learning enables online learning where data is continuously streamed from sensors. Professor Rosing’s team demonstrates the value of HDC in communication networks, Internet of Things systems, and distributed machine learning architectures.

Professor Tim Kraska, Massachusetts Institute of Technology, USA
Instance-Optimized Database Systems

This project broadly explores the use of AI for optimizing large-scale data systems and enterprise applications. Professor Kraska’s team has shown how to enhance core components of a data system (e.g., index structures, sorting algorithms, query optimizers) by learning models of data distributions and query workloads. This research has led to the creation of SageDB – an instance-optimized database system that self-adjusts for a given data, workload, and hardware setting. 

Professor Marian Verhelst and Dr. Wannes Meert, Katholieke Universiteit (KU) Leuven, Belgium
Professor Guy Van den Broeck, University of California Los Angeles, USA 

Probabilistic Computing

This project explores a novel AI approach that captures the probabilistic and logical relationships between variables of interest to enable efficient learning and tractable inference. This vertically integrated research across the probabilistic computing stack has involved aspects such as novel representations, algorithms, probabilistic programming languages, and hardware integrated circuit implementation with a focus on improving energy efficiency and throughput. The research also enables efficient quantum simulation via probabilistic computing.

Professor Chinedum Osuji, University of Pennsylvania, USA
Patterning at Nano‐Length Scales by Directed Assembly

This project focuses on patterning at sub-10 nm length scales by synergizing the concept of block-copolymer-based bottom-up self-assembly with novel liquid-crystal molecular chemistry. This research will enable a new class of novel materials that could play a key role in continued scaling of device structures.

Professors Justine Sherry, Vyas Sekar, and James C. Hoe, Carnegie Mellon University, USA
Pigasus: FPGA-Accelerated Intrusion Detection and Prevention System

The Pigasus Intrusion Detection and Prevention System inspects 100k+ concurrent connections against 10k+ rules. On this stateful and irregular computation, Pigasus’ FPGA-first architecture efficiently achieves 100 Gbps within a single server form factor by handling common-case processing in the FPGA SmartNIC, offloading to only a few CPU cores in rare cases. The Crossroads Research Center has open-sourced Pigasus and hosts a developer’s group.

Professor Ron Dror, Stanford University, USA
Geometric Deep Learning of RNA Structure

This project focuses on a machine learning method that enables identification of accurate structural models of RNA. RNA performs a wide range of key cellular functions. Predicting the three-dimensional structures of RNA molecules is critical for understanding their function. The research team designed the Atomic Rotationally Equivariant Scorer (ARES), and the team’s work was featured on the cover of the August 27, 2021 issue of Science.