Sungjae Lee Email and Phone Number

• Develop inference system for LLM, HyperCLOVA X.• Research optimization techniques for serving LLM.

Serving System Development for LLM, HyperCLOVA• Led the development of a high-performance serving system for Naver's LLM, HyperCLOVA.• Proficiently utilized transformer frameworks (Megatron-LM, FasterTransformer) for robust model deployment.• Applied optimization techniques for efficient model serving.Engineering ML Training Platform on Kubernetes• Contributed to ML training platform development and operations.• Designed public APIs for seamless interaction and enhanced usability.

• Implemented deep learning framework and benchmark.• Explored methods for accelerating deep learning, including pipelining and pruning.• Devised a memory system for server workloads.