Job Description

Position SummaryLooking for a professional with 8+ years of experience in implementation of AI (need not necessarily be in wireless domain) specially on GPU based platforms.Role and Responsibilities1. Design and develop generic AI acceleration framework for GPUs, CPUs and NPUs.2. Prune and optimize trained AI models of telco use cases.3. Optimize Neural Network libraries to best adopt to underlying compute platform.4. Implement and verify the generic AI frameowrk and optimized models. Profile on various compute patforms5. Define functional and performance test cases against requirement and design choices.6. Contribute to Samsung Intellectual Property (patents, tech papers) from the experiments and analysis done from the project.7. Work together with internal members of vRAN, L2 protocol stack and embedded system SW for product integration.Skills and Qualifications1. 8+ years of experience in low level embedded SW development: such as architecting SW as per CPU architectures and knowledge of Assembly level debugging.2. 5+ years of experience in in AI/ML optimization: both in terms of AI model pruning and optimizing Neural Network library.3. 5+ years of experience in training AI/ML models, fine-tuning using Tensorflow or Pytorch and integration of trained models into relevant product SW.4. 3+ years of experience in NVIDIA GPU platforms and AI/ML ecosystem: such as CUDA programming, DOCA SW framework.5. Expertise knowledge in programming vector processing SW on SIMD processors and GPUs.5. Prficient knowledge in hosting LLMs on GPUs. Knowledge of routing AI execution jobs across CPUs and GPUs.6. Good understanding of the High speed serial interfaces such as Ethernet, PCIe.7. Hands-on experience in SW integration of multi-core, multi-module multi-developer environment.Advantageous to have:- Experience in telecommunication and Radio Access Network SW development- Hands-on AVX instructions and optimizations on the assembly level- Hands-on GPU programming and/or SIMD programming for modem Physical layerKeywords:[GPU] [CUDA programming] [AI ML acceleration] [AI ML optimization] [Nueral Network] [Assembly][SIMD] [Multi-tanancy][Multi-tenent]

• Please visit to see Privacy Policy, which defaults according to your location. You can change Country/Language at the bottom of the page. If you are European Economic Resident, please click .

Samsung