Summary
Junhong Kim is a Generative AI Team Lead with a Ph.D. in Machine Learning and nine years of experience building production-grade AI systems for enterprise manufacturing. At SK Siltron he architected and shipped a fully in-house generative AI platform—combining LLMs, VLMs, custom multi-agent RAG, layout-aware visual parsing, and optimized low-latency inference—to turn fragmented corporate PDFs, slides, and transcripts into decision-ready knowledge services. Previously at Samsung Electronics he led domain-specific RAG and LLM optimization for semiconductor use cases, including real-time EUV overlay map retrieval and AI-driven etch recipe advisors validated in production. He’s skilled at bridging infrastructure, reasoning, and data—designing everything from training pipelines for vision retrievers to distributed H200 GPU clusters and prompt-compression strategies for cost-efficient inference. A detail that’s not obvious: he deliberately avoids off-the-shelf frameworks, preferring bespoke architectures that tightly integrate vision and language for high-stakes, explainable enterprise workflows.
8 years of coding experience
Doctor of Philosophy - PhD, Machine learning, Industrial Engineering, Doctor of Philosophy - PhD, Machine learning, Industrial Engineering at 고려대학교