Jun Kim
Building high-performance AI tools and data infrastructure for local execution, specifically optimized for Apple Silicon hardware.
At a Glance
- AI Developers
- Apple Silicon users
- Open-source contributors
- Data Engineers
AI Tools by Jun Kim
(1)oMLX
LLM Inference Server for Apple Silicon
Discussions
No discussions yet
Be the first to start a discussion about Jun Kim
Latest News
oMLX v0.4.5.dev1 released with major speedups for GLM-5.2 and MiniMax M3
oMLX v0.4.4 stable release addresses macOS 27 compatibility issues
Native MarkItDown support added to oMLX for document processing
oMLX crosses 17,000 GitHub stars within months of launch
Products & Services
An open-source LLM inference server for Apple Silicon with continuous batching, SSD KV caching, and a native macOS menu bar app.
An AI-powered Chromium browser designed for productivity and live conversation intelligence.
Market Position
oMLX competes with local LLM servers like Ollama and LM Studio, distinguishing itself through deep optimization for Apple Silicon (MLX), native macOS application design, and advanced features like continuous batching and SSD-based KV cache persistence which are critical for coding agents.
Leadership
Founders
Jun Kim (Heejun Kim)
Global business planning at a major Korean food company (data infrastructure design); Founding Engineer & Backend Team Lead at Modhaus; Co-founder & CEO at Aside (YC F25).
Executive Team
Jun Kim (Heejun Kim)
Creator & Lead Developer
Background in global business planning and data infrastructure; former Founding Engineer at Modhaus; CEO of Aside.
Founding Story
Jun Kim started oMLX to address the limitations of existing local LLM tools on macOS, particularly focusing on solving KV cache invalidation issues for coding agents and enabling massive context through SSD-tiered caching.
Business Model
Revenue Model
Open Source (Apache 2.0)
Target Markets
- AI Developers
- Apple Silicon users
- Open-source contributors
- Data Engineers
- Local LLM inference
- Coding agent backends
- Multimodal document processing
- Low-latency local AI assistants