Intel is proud to announce the recipients of its 2024 Outstanding Researcher Awards (ORAs), honoring 10 distinguished academic researchers whose work is making a significant impact on the future of technology. This annual award program celebrates exceptional achievements made through Intel-sponsored university research that supports Intel’s mission to create world-changing technology that enriches the lives of every person on Earth.
The ORAs are a key initiative of Intel’s Corporate Research Council, which fosters deep collaborations with leading universities, research centers, and ecosystem partners. The 2024 awardees are advancing innovation across a wide range of areas, including AI-driven materials design, scalable computing optimization, domain-specific compiler technologies, cutting-edge accelerator architectures, advanced metrology, field-programmable gate arrays (FPGAs), next-generation semiconductor devices, and privacy-enhancing knowledge sharing. These research partnerships are helping drive the transformation of today’s computing into the breakthroughs of tomorrow.
The winners of Intel’s 2024 Outstanding Researcher Awards are:
Professor Amit Trivedi, University of Illinois, Chicago, USA
Scalable Multibit Precision In-Memory Deep Neural Network Processing by Co-Designing Neural Network Operators
The collaborative work on MC-CIM: Compute-in-Memory with Monte-Carlo Dropouts for Bayesian Edge Intelligence introduced a novel framework for in-memory Bayesian inference by embedding Monte Carlo dropout techniques directly within compute-in-memory architectures, enabling uncertainty-aware, energy-efficient AI at the edge. The research team explored a frequency domain learning approach using binarized Walsh-Hadamard transforms to significantly reduce the necessary parameters for deep neural networks, making compute in static random-access memory (SRAM) especially advantageous. They proposed a novel approach to energy-efficient acceleration of frequency-domain neural networks using analog-domain frequency-based tensor transformations, achieving more compact cells and enabling adaptive stitching of cells column-wise and row-wise for a high degree of parallelism.
Professor Anwar Hithnawi, ETH Zurich, Switzerland
Robust and Automated Generation of Systems for Fully Homomorphic Encryption
Fully homomorphic encryption (FHE) requires compilers to translate general-purpose computations into cryptographic operations. The Privacy-Preserving Systems Lab, led by Anwar Hithnawi, developed a first-of-its-kind end-to-end compiler design that addresses all stages of the FHE development workflow. In addition, the team explored the use of FHE in adversarial settings, systematizing approaches to integrity protection and evaluating practical techniques for integrating zero-knowledge proofs.
Professor Azad Naeemi, Georgia Institute of Technology, USA
Compact Physical Models and Circuit Design for Ferroelectric and Antiferroelectric Devices
This project established a general framework to build compact circuit models for ferroelectric capacitors and extended the models to multi-phase hafnia-based ferroelectric and anti-ferroelectric capacitors for wider experimental data coverage. The team also performed comprehensive analyses of ferroelectric random-access memory (FeRAM) arrays and peripheral circuitry designs using the compact models for emerging memory benchmarking.
Professor Benjamin Tan, University of Calgary, Canada
Development of a Common Weakness Enumeration Exploration Toolkit for Register Transfer Level
The research team transferred a register transfer level (RTL) static analysis scanner to detect hardware common weakness enumerations and demonstrated feasibility for open-source chip designs. They also transferred an RTL bug repair tool assisted by a large language model and developed a proof of concept for determining root cause failures in digital hardware design processes.
Professor Darrell Schlom, Cornell University, USA
Material Stack Engineering of All-Oxide Ferroelectric Field-Effect Transistor
The research team eliminated a buffer layer and decreased the thickness to successfully deposit La-BaSnO3 with high mobility and targeting low carrier concentration. This enabled the fabrication of an all-perovskite ferroelectric field effect transistor (FeFET) and meeting established performance metrics.
Professor Dimitrios Skarlatos, Carnegie Mellon University, USA
Rebuilding Virtual Memory for Heterogeneous Architectures
The research team explored address translation overheads from limited memory contiguity in the datacenter. They delivered techniques to increase contiguity, yielding significant performance improvements, and have shared their innovation with the community through Linux kernel patches. The team applied lightweight machine learning to address translation and developed Learned Virtual Memory that delivers performance within 1% of an ideal page table.
Professor Huolin Xin, University of California, Irvine, USA
AI-Enabled High-Throughput High-Resolution Electron Tomography for High-Throughput Failure Analysis and Atom Probe Tomography Geometric Correction
The research team has developed a novel approach for fast 3D tomography characterization using artificial intelligence. The project delivered a significant improvement in data acquisition speed compared to prior methods used. This significantly enhances throughput and helps to address 3D metrology challenges, including complicated nanoribbon and power via architectures. The capability will be instrumental for power-performance enhancements, as well as yield and reliability improvements.
Professor John Heron, University of Michigan, Ann Arbor, USA
Enabling Robust and Energy Efficient Magneto-Electric Spin-Orbit Devices Using High Entropy Oxides with Large Spin Hall Resistivity
The research team demonstrated ultrafast switching of La-doped BiFeO3 ferroelectric capacitors, developed novel metrologies to measure polarization dynamics at nanoscale, demonstrated modeling frameworks to understand the effect of key physical processes such as domain nucleation, growth, and circuit limits on the switching process, and determined a new regime of energy-delay scaling behavior relevant for computing technologies. Furthermore, the researchers developed novel materials critical for accelerating magneto-electric spin-orbit (MESO) device development to deliver target specifications, such as high entropy perovskite oxides with large spin Hall efficiency and resistivity as well as double perovskite ferromagnet layers epitaxially compatible with La-doped BiFeO3.
Professor Pablo de Oliveira Castro, University of Versailles Saint Quentin, France
Leveraging Statistical Learning to Support Numerical Software Optimization
With the Machine-Learning for Kernel Accuracy and Performance Studies (MLKAPS) framework, the research team proposed a fully automated and scalable autotuning framework based on machine learning to improve productivity in tuning Intel Math Kernel Library kernels. This increases software performance, shortens time to market, and enables tailoring libraries for specific needs.
Professor Ronald Reano, Ohio State University, USA
Quantum Curriculum Using the Intel® Quantum Software Development Kit
Professor Reano and his team developed an academic curriculum based on the Intel Quantum Software Development Kit (IQSDK) across multiple university departments. The coursework targeted a wide range of learners, from undergraduates to graduate students and already practicing engineers. This type of shareable curriculum enables future developers to contribute to and participate in a growing quantum computing ecosystem.