Lisa Maile

Professor at Eindhoven University of Technology

Eindhoven University of Technology
Country flagNetherlands

Research Interests

ApplyKite TwinLearns as you apply

Make your next move.

Prepare your application

Review drafts before you send them.

Explore related searches

About

Lisa Maile is an Assistant Professor at Eindhoven University of Technology in the Netherlands. Her research focuses on real-time latency prediction in dynamic 5G communication, network calculus, and routing algorithms for ultra-reliable applications. She has contributed to significant works on delay-guaranteeing admission control for time-sensitive networking. With a robust portfolio of articles, she actively addresses the challenges in networking and communication technologies.

Recent Grants

Grant: Close

Platforms for Information Technology

Open Date:

Close Date:

Positions (1)

Publisher
source

Lisa Maile

University Name
.

Eindhoven University of Technology

This PhD project investigates how communication networks can be made faster and more energy-efficient for large language model inference. When inference is distributed across multiple GPUs, the devices must exchange intermediate results continuously. Network congestion, communication overhead, and buffer requirements can therefore limit system performance and increase energy consumption. As part of the Communication-Optimized Inference in Large Language Models (COIL) project, the doctoral researcher will develop analytical models and orchestration methods for networks connecting GPUs in AI clusters. The research will use Network Calculus to create fast, composable models that capture interactions between inference traffic, network architecture, flow control, and congestion feedback. These models will provide bounds on communication latency and buffer requirements and will be complemented by measurement-calibrated estimates of energy use. The project will examine how network topology, link bandwidth, and parallelization strategies influence performance and energy consumption. The researcher will develop tools to help cluster designers assess configurations and will investigate model-informed runtime decisions such as workload placement, prioritization, routing, and resource reservation. A prototype will demonstrate how these decisions can improve the operation of an existing network. Experimental validation is central to the project. The researcher will work with AI infrastructure measurements and conduct reproducible experiments on programmable networking testbeds using technologies including Ethernet-based AI networks, Ultra Ethernet mechanisms, P4-programmable switches, and SmartNICs. Results will be evaluated on real AI infrastructure in collaboration with SURF. The work combines mathematical theory, software development, network experimentation, and practical AI infrastructure research, with opportunities to share findings with networking standardization communities and release tools and experimental results openly. The main supervisor is Professor Lisa Maile, leader of the Network Modeling and Design (NoMaD) lab in the Electronic Systems group of the Department of Electrical Engineering at Eindhoven University of Technology. The research will connect with the Advanced Models by Open Research and Engineering Lab in Computer Science and the Model-Based Design Lab in Electrical Engineering. The doctoral researcher will also collaborate with Professor Georg Carle and the Network Architectures and Services group at the Technical University of Munich, spending six months in Munich to gain hands-on experience with programmable network hardware and reproducible experimental infrastructure. NoMaD develops mathematical models, tools, and protocols for reliable and efficient communication networks, combining formal analysis with simulation and hardware experiments. The broader Electronic Systems group provides an interdisciplinary environment spanning system architectures, software, digital circuits, predictable performance, timing, reliability, and energy consumption. The project is situated in Eindhoven’s international Brainport technology region and in TU/e’s collaborative research environment focused on high-tech engineering and socially relevant applications.

3 days ago

Articles (4)

Year:

2025

Year:

2024

Year:

2023

Year:

2022