research Near 1-Bit Ultra-Low-Bit Quantization for Trillion-Scale LLMs Research on efficient inference for trillion-scale language models at the Flexible Machine Learning Lab, with Samsung Electronics. competition ZipJiGi Excellence Award (KT Award), Top 1.33% — 2026 KB Financial AI Challenge, hosted by KB Kookmin Bank. NVIDIA Nemotron 3 30B Model LoRA Adapter Development LoRA adapter development for the 2026 NVIDIA Nemotron3 Reasoning Challenge. LG AI Research EXAONE 4 1.2B Model Quantization Top 3% in the semifinal round, advanced to the final round — 2026 LG AI Research EXAONE Compression Competition. build Amplify A question-driven collective learning platform that reconnects classrooms in the age of AI. Easy Coach A computer-vision-based posture correction tool that analyzes and coaches exercise form.
