PhD Position (f/m/d) - Memory-Centric NPU Architectures and Emerging Memory Technologies for Edge AI
As AI models continue to grow in size and complexity, memory subsystems are emerging as a fundamental bottleneck in modern AI accelerators. The efficiency of future NPU-based systems will increasingly depend on how data is stored, moved, and processed across the memory hierarchy.
We are looking for a PhD candidate to work on Memory-Centric NPU Architectures and Emerging Memory Technologies for Edge AI. This PhD project focuses on exploring memory-centric AI accelerator architectures that integrate advanced NPUs with emerging memory technologies such as eDRAM and next-generation embedded memories. The goal is to develop innovative hardware/software co-design methodologies that improve performance, power efficiency, memory utilization, and scalability for future edge AI and embedded Generative AI systems.
In this role, you will focus on the following research directions:
- Memory-Centric NPU Architecture Design: Investigate memory-centric NPU subsystem architectures that improve performance, scalability, and energy efficiency for future AI workloads.
- Emerging Memory Technologies for AI Acceleration: Evaluate and model emerging memory technologies such as eDRAM, GCRAM, and other next-generation embedded memories for future AI accelerators. Analyze their impact on performance, power consumption, silicon area, reliability, and scalability.
- Hardware/Software Co-Optimization for AI Workloads: Develop hardware/software co-optimization techniques for efficient execution of AI workloads on NPU-based systems.
- Design Space Exploration and Performance Modeling: Develop architecture exploration frameworks that enable rapid evaluation of future AI subsystem designs and quantify trade-offs across performance, power, area, memory utilization, and workload characteristics.
You will work at the intersection of computer architecture, semiconductor memory systems, AI accelerators, embedded systems, and machine learning, helping shape the next generation of NXP AI processors.
In this PhD, you will:
- Conduct research on next-generation NPU subsystem architectures
- Develop architectural modeling and simulation frameworks
- Investigate the integration of emerging memory technologies into AI accelerators
- Build performance, power, and area (PPA) evaluation methodologies for AI hardware design
- Evaluate real-world AI workloads including CNNs, Transformers, LLMs, VLMs, and multimodal AI systems
- Analyze trade-offs across performance, power, memory bandwidth, silicon area, scalability, and reliability
- Publish results in leading conferences and journals
- Collaborate with experts in semiconductor design, system architecture, AI acceleration, and embedded AI
Your Profile:
- A Master's degree in Electrical Engineering, Computer Engineering, Computer Science, Microelectronics, Embedded Systems, or a related field
- Strong background in one or more of the following:
- computer architecture
- digital design
- embedded systems
- semiconductor memory systems
- AI accelerators and NPUs
- hardware/software co-design
- digital hardware design
- Experience with Python and C/C++
- Knowledge of computer architecture simulation, performance modeling, or hardware design methodologies
- Strong analytical and problem-solving skills
- Very good written and spoken English
The following experiences would be considered a plus:
- AI accelerator and NPU architecture
- RTL development using Verilog/SystemVerilog/VHDL
- FPGA prototyping
- Computer architecture simulators
- Memory hierarchy optimization
- eDRAM, GCRAM, MRAM, ReRAM, or CXL-based memory systems
What you can expect:
At NXP, you will work on research that directly contributes to future generations of AI processing hardware. You will have access to state-of-the-art NPU architectures, architecture simulation frameworks, advanced semiconductor design methodologies, and emerging memory technologies currently being investigated for future AI products. You will collaborate with experts in AI hardware architecture, Digital IP design, memory systems, and machine learning while contributing to technologies that can influence future NXP Edge AI and Generative AI processors.
This PhD offers a unique opportunity to combine cutting-edge academic research with high industrial relevance and to contribute directly to the development of next-generation AI hardware architectures.
Please note: The successful candidate may/will be responsible for security related tasks. The assignment may/will be in scope of security certifications, therefore a conscious and reliable way of working is necessary.
#LI-3da9