At Tether, I research on-device Small Language Models and optimize them for constrained compute, memory, and latency budgets. I also design benchmarks to measure accuracy, latency, memory footprint, and power efficiency.
At Naver, I developed production-grade agentic AI for search, shopping, and enterprise services, and led LLM post-training workflows for domain-specific tasks. My work included AI Briefing, a real-time RAG summarization service, and Search Big Bench, a benchmark for LLM performance in real-world search scenarios.
