Senior Staff Research Scientist, On Device/Efficient AI/LLM

Lab Summary:  The Samsung AI Center (SAIC) within Samsung Research America (SRA) leads at the forefront of innovation in creating intelligent and interactive machines building upon an ecosystem that is user-centric rather than device-centric. The success of AI will depend on how well devices understand their users – and how well devices empower users.  SAIC takes on grand scientific and engineering challenges in machine intelligence and actively contributes to the international research community through scientific publications and presentations in major conferences and journals in research areas of Computer Vision, HCI, Contextual/multi-modal modeling, NLP, Speech Recognition, Dialogue, and Machine and Deep Learning. The SAIC is a key part of Samsung’s global R&D effort and aims to have influence on future Samsung products reaching hundreds of millions of users worldwide.

 Position Summary: We are looking for candidates with machine learning and deep learning background, in the field of on-device and efficient AI. You will work with a team of research scientists and engineers tackling real-world problems involving Samsung’s Artificial Intelligence initiatives. You will be involved in very promising team projects with talented people at Samsung. You will benefit a lot by working in a fun and creative environment. The AI research center is a key part of Samsung’s global R&D effort and aims to have influence on future Samsung products reaching hundreds of millions of users worldwide.

Position Responsibilities:

• Design, develop and implement novel efficient model architectures of large foundation models (e.g., LLM, LVM, etc.) for various applications including language, vision, audio, sensor data, etc.
• Develop and implement efficient model training algorithms, such as LoRa, DoRA, etc.
• Develop and implement efficient inference algorithms, such as speculative decoding, and model parallelism, etc.
• Deploy and optimize efficient models on edge devices (e.g., Samsung phone GPU/NPU).
• Generate creative solutions (patents) and publish research results in top conferences (papers).

Required Skills:

• Ph.D. in C.S., EE or equivalent combination of education, training, and experience
• 15 years of experience with AI related Research.
• Bi-lingual Korean language is highly preferred.
• Expertise in the state-of-the-art model architectures, including Transformer, Mamba, Liquid neural networks, etc.
• Experience in efficient model training algorithms and model design, including LoRa, DoRA, Flash Attention, FlashAttention-2, etc.
• Experience in efficient model inference algorithms, including speculative decoding, KV caching, etc.
• Experience in model compression techniques (e.g., pruning, quantization, knowledge distillation, SVD, etc.).
• Experience in developing and optimizing deep learning models (e.g., Transformer, Mamba, liquid) on edge devices (e.g., GPU/NPU).
• Proficiency in on-device neural network libraries (e.g., llama.cpp, MLC LLM, etc.);
• Experience in efficient hardware accelerator design for neural computing on mobile devices.
• Proven track record of research/publications on machine learning and artificial intelligence field (NeurIPS, ICLR, ICML, AAAI, IJCAI, CVPR, ACL, etc.)