Chanwoo Park's Blog

A.X K1: SK Telecom's 519B Parameter Reasoning Model

SK Telecom releases A.X K1, a 519B parameter MoE LLM that outperforms DeepSeek-V3.1 in math and coding. As an intern, I contributed to enhancing its reasoning capabilities.

3 min read · January 07, 2026

2026 · llm moe reasoning skt internship · research
Building My Personal AI Infrastructure with MCP Servers

How I built a personal AI infrastructure using K3s, MCP servers, and GitOps to automate my workflows

7 min read · December 14, 2025

2025 · ai infrastructure mcp kubernetes gitops · infrastructure
Llama-Thunder-LLM: Building a Korean-Specialized Large Language Model

Our research on developing Korean-specialized LLM from Llama, featuring Thunder-Tok tokenizer and Korean benchmarks

2 min read · July 02, 2025

2025 · llm korean nlp research · research