Sort by
Refine Your Search
-
Listed
-
Employer
-
Field
-
model in collaboration with partner institutions such as the German Climate Computing Center (DKRZ) and German Weather Service (DWD), including GPU porting. They will perform production runs of ICON and
-
, mathematics or any related field. What we offer State of the art on-site high performance/GPU compute facilities Competitive research in an inspiring, world-class environment A wide range of offers to help you
-
containers (Docker/Singularity/Podman/Kubernetes). Experience with Ethernet, InfiniBand, RDMA network technologies. CPU/GPU/memory/RAID/storage/Data Center technologies. Knowledge of current technological
-
physics, mathematics or any related field. What we offer State of the art on-site high performance/GPU compute facilities Competitive research in an inspiring, world-class environment A wide range of offers
-
Your profile: Preferably a doctoral degree, but MSc are also encouraged to apply Expert knowledge in one or several of the following High Performance Computing GPU computing Array Computing with JAX A
-
commonly used on Unix systems. Additional languages or experience with libraries for utilizing GPU hardware efficiently, e.g., CUDA, are a plus. Experience in AI programming with, e.g., PyTorch(-DDP
-
benchmark them with a realistic case study. The main focus of the project can develop either more in the mathematical theory of MCMC, the implementation of code for the Jülich supercomputers (GPU/CPU
-
, PyTorch) for ML applications, training, evaluation, and deployment of models Use of GPU-based servers and modern IT infrastructure for training and inference Application of classical ML methods (e.g
-
, including Large Language Models (LLMs), agent-based systems, and Retrieval-Augmented Generation (RAG). Practical expertise in training and optimizing neural networks on high-performance (GPU-enabled
-
. Additional languages or experience with libraries for utilizing GPU hardware efficiently, e.g., CUDA, are a plus. Experience in AI programming with, e.g., PyTorch(-DDP), Horovod, or DeepSpeed, and in