Code Optimizer

Fujitsu Limited; Singapore; Presencial; SGD 120.000 - 180.000; Jornada completa

Descripción de la vacante

Fujitsu Limited in Singapore seeks an experienced HPC/AI performance engineer to plan and execute optimization projects. You will engage with customers to analyze and parallelize AI workloads, profile bottlenecks, and propose GPU-accelerated solutions.

The role requires strong C/C++, Python, MPI/OpenMP skills and hands-on experience with CUDA, PyTorch, and TensorFlow in distributed environments. Knowledge of climate or aeronautics simulations is a plus.

Formación

• 4+ years of experience in application development and performance optimization.
• Experience with numerical model parallelization.
• Strong programming and parallel computing skills.
• Familiarity with HPC and AI architectures.
• Experience with climate or aeronautics simulations is a plus.

Responsabilidades

• Plan and execute HPC and AI application performance tuning and optimization activities per project schedules.
• Engage with customers to analyse, parallelize and optimize HPC and AI applications for performance and scalability.
• Profile and analyse HPC/AI workloads to identify bottlenecks and propose optimization approaches.
• Optimize GPU-accelerated and distributed AI workloads using CUDA, PyTorch, and distributed frameworks.
• Conduct benchmarking and scalability analyses to establish baselines and evaluate improvements.
• Serve as a subject matter expert in HPC and AI optimization to develop optimized solutions.

Conocimientos

Fortran

C

C++

Python

MPI

OpenMP

UNIX/Linux

CUDA

PyTorch

TensorFlow

Distributed training

Herramientas

Linaro Forge

TotalView

NVIDIA Nsight Systems

NVIDIA Nsight Compute

Descripción del empleo

Location Flexibility: Primary Location Only

Key Responsibilities:

• Plan and execute HPC and AI application performance tuning and optimization activities in accordance with agreed project schedules and objectives.
• Engage with customers and end users to analyse, parallelize and optimize HPC and AI applications, improving performance, scalability and efficient use of computing resources.
• Profile and analyse HPC and AI workloads to identify performance bottlenecks across CPU, GPU, memory, I/O and inter-node communication, and recommend appropriate optimization approaches.
• Optimize GPU-accelerated and distributed AI workloads, including multi-GPU and multi-node workloads using technologies such as CUDA, PyTorch and distributed training frameworks.
• Conduct application benchmarking and scalability analysis to establish performance baselines, evaluate optimization improvements and support efficient utilisation of HPC resources.
• Serve as a subject matter expert in HPC and AI application optimization, applying strong technical expertise and industry knowledge to develop optimized solutions that support customer requirements and business objectives.

Job Requirements:

• Solid understanding of HPC and AI architectures, including CPU/GPU-based computing, parallel computing and distributed workloads.
• Minimum 4 years of relevant experience in application development, performance optimization and numerical model parallelization.
• Strong programming and parallel computing skills in Fortran, C, C++, Python, MPI and OpenMP.
• Strong working knowledge of UNIX/Linux operating systems and HPC application environments.
• Knowledge and experience with numerical simulation applications, such as climate research, weather forecasting, computational science, engineering or aeronautics simulation.
• Proficient in using performance analysis, profiling, compiler analysis and debugging tools, such as Linaro Forge, TotalView, NVIDIA Nsight Systems/Nsight Compute, or equivalent tools.
• Knowledge of GPU computing and CUDA, including profiling and optimization of GPU-accelerated applications.
• Experience with AI/ML frameworks such as PyTorch and/or TensorFlow and running AI workloads in GPU-based computing environments.
• Knowledge of distributed AI training and multi-GPU/multi-node workloads, including technologies such as PyTorch Distributed and Distributed Data Parallel (DDP).
• Strong communication and presentation skills, with the ability to develop and deliver technical workshops, training and knowledge-sharing sessions to HPC and AI users.