EveryDev.ai
Subscribe
Home
Developers

4,896+ AI companies

  • Radar
  • Trending
AI Tools by Topic
  • AI Coding Assistants
  • Agent Frameworks
  • MCP Servers
  • AI Prompt Tools
  • Vibe Coding Tools
  • AI Design Tools
  • AI Database Tools
  • AI Website Builders
  • AI Testing Tools
  • LLM Evaluations
Follow Us
  • X / Twitter
  • LinkedIn
  • Reddit
  • Discord
  • Threads
  • Bluesky
  • Mastodon
  • YouTube
  • GitHub
  • Instagram
Get Started
  • Users
  • Rate Tools
  • About
  • Editorial Standards
  • Corrections & Disclosures
  • Community Guidelines
  • Advertise
  • Contact Us
  • Newsletter
  • Submit a Tool
  • Start a Discussion
  • Write A Blog
  • Share A Build
  • Terms of Service
  • Privacy Policy
Explore with AI
  • ChatGPT
  • Gemini
  • Claude
  • Grok
  • Perplexity
Agent Experience
  • llms.txt
Theme
With AI, Everyone is a Dev. EveryDev.ai ©
    1. Home
    2. Developers
    3. Raullen Chai

    Raullen Chai

    Raullen Chai is an individual developer and creator, not a company. His public work focuses on building open intelligence for the physical world, including local AI inference on Apple Silicon and an ESP32 voice-assistant project.

    Visit Website

    At a Glance

    2Tools Listed
    3Products
    119Tool Views
    10Capabilities
    Discussions
    San Francisco, CaliforniaHeadquarters
    2017Est.
    100Employees
    $80000000Raised
    Focus Areas
    AI Coding Assistants
    Speech Recognition
    Conversational Agents
    Voice Assistant
    Connect
    Latest News
    The Last Shift: Qwen Image and LTX 2.5 Video with ComfyUI on a MacOct 11, 2026
    Make Imaginary Travel Posters Locally with ComfyUI and Rapid-MLXOct 10, 2026
    Markets
    • Apple Silicon Mac users
    • Developers building coding agents and local AI applications
    • Privacy-conscious users who want inference to stay on-device
    • MLX and open-source model researchers
    • +2 more

    AI Tools by Raullen Chai

    (2)
    View esp32-thinks
    esp32-thinks tool icon

    esp32-thinks

    ESP32 Voice Assistant Firmware

    Voice AssistantConversational AISpeech Recognition
    View Rapid-MLX
    Rapid-MLX tool icon

    Rapid-MLX

    MLX LLM Server for Apple Silicon

    Local InferenceAI Coding Asst.LLM Orchestration

    Discussions

    No discussions yet

    Be the first to start a discussion about Raullen Chai

    Latest News

    10/11/2026

    The Last Shift: Qwen Image and LTX 2.5 Video with ComfyUI on a Mac

    rapidmlx.com
    10/10/2026

    Make Imaginary Travel Posters Locally with ComfyUI and Rapid-MLX

    rapidmlx.com
    10/06/2026

    Use a Local Model from Python on Your Mac

    rapidmlx.com
    10/06/2026

    Run Cline with a Local Model on Your Mac

    rapidmlx.com

    Products & Services

    3
    Rapid-MLX
    July 24, 2026 (0.11.0 release documented)

    Open-source Apache-2.0 local LLM inference server and Mac desktop/CLI engine for Apple Silicon. It serves MLX models through OpenAI- and Anthropic-compatible APIs, supports coding agents and tool calling, and runs without an account or cloud inference.

    esp32-thinks
    October 12, 2026 (v0.1.0 initial release documented in repository history)

    A multi-model personal AI voice assistant for ESP32-S3/Muse Gadget hardware, especially the Seeed SenseCAP Watcher W1-B. It combines wake-word or knob interaction with configurable OpenAI-compatible speech-to-text, chat and text-to-speech endpoints, streaming replies and an optional openmuse agent.

    openmuse
    October 2026

    An optional agent component included in esp32-thinks that keeps extra context such as location, memory and tools on the user's own machine and exposes an OpenAI-compatible chat endpoint for the device.

    Market Position

    Rapid-MLX positions itself as a faster, agent-focused local inference alternative to Apple's mlx-lm and as an alternative to oMLX, LM Studio, Ollama and llama.cpp on Apple Silicon. Its differentiation is the combination of MLX-native performance, OpenAI/Anthropic API compatibility, schema-valid tool calling, MCP support, multimodal runtimes and local-first privacy. The official comparison pages report direct same-model comparisons against oMLX and LM Studio, while the blog also benchmarks against Ollama and llama.cpp.

    Leadership

    Founders

    RC

    Raullen Chai

    Cryptographer and software engineer with a Ph.D. in Electrical and Computer Engineering from the University of Waterloo (2011). Previously worked as a Member of Technical Staff at Oracle, Senior Software Engineer/Tech Lead at Google, and Lead of Crypto R&D and Engineering Security at Uber. He is also identified by CryptoSlate and IoTeX as a co-founder and former CEO of IoTeX and MachineFi Lab.

    Executive Team

    RC

    Raullen Chai

    CEO

    Cryptography expert, former Uber and Google security leader.

    AB

    Aaron Basi

    Head of Product

    Focuses on product strategy for DePIN infrastructure.

    Board of Directors

    JL
    Jack Lee
    Advisor (Founding Managing Partner at HCM Capital/Foxconn)
    RP
    Robert Parker
    Advisor (Former CTO of Samsung SmartThings)
    MA
    Mukul Agarwal
    Advisor (Founder of BootUP Ventures)

    Founding Story

    In a Show HN post, Chai said he built Rapid-MLX to run coding agents locally on Apple Silicon after repeatedly encountering two problems: models failing at structured tool calling and existing MLX servers being slow. The project's stated vision is a fast, local, privacy-preserving backend that works with agent workflows rather than a cloud API.

    Business Model

    Revenue Model

    The publicly documented projects are free and open source. Rapid-MLX is Apache-2.0 licensed, and the website says the engine and Desktop app are free with local models running on the user's Mac; no paid API or cloud-inference model is described.

    Pricing Tiers

    Rapid-MLX / Desktop
    Free

    Free download and open-source engine; macOS 14+ and Apple Silicon are required for the Desktop download described on the official site.

    Private

    Target Markets

    Industries & Segments
    • Apple Silicon Mac users
    • Developers building coding agents and local AI applications
    • Privacy-conscious users who want inference to stay on-device
    • MLX and open-source model researchers
    • ESP32/Muse Gadget makers and edge-AI developers
    • Users wanting local speech, vision and multimodal AI
    Use Cases
    • Private local LLM inference and chat on Mac computers
    • Local coding agents and pair-programming workflows
    • Connecting Claude Code, Cursor, Cline, Continue, Goose, OpenCode, Pi, Aider and Codex CLI to local models
    • Local MCP tool use and agent workflows
    • Running local speech, vision and multimodal models
    • Building an offline-first or privacy-preserving ESP32 voice assistant
    Notable Customers
    • Samsung Next
    • Bosch
    • Helium
    • Polygon

    Quick Facts

    Headquarters
    San Francisco, California, United States
    Founded
    2017
    Entity Type
    Foundation
    Employees
    100
    Total Funding
    $80,000,000+
    Investors
    Borderless Capital, Amber Group
    Office Locations
    Menlo Park
    Singapore
    Shenzhen
    Tortola

    Funding History

    Private Sale$30,000,000
    Feb 2018
    Draper Dragon
    Kenetic Capital
    Digital Currency Group
    Ecosystem Fund$50,000,000
    April 2024
    Borderless Capital
    Amber Group
    Foresight Ventures

    History & Milestones

    2026-10-19

    The esp32-thinks repository recorded the latest commit adding an ESP32 community survey and feature ideas, while the repository's release work included v0.2.10/v0.2.11 changes and the openmuse integration.

    2026-10-12

    The Rapid-MLX repository homepage described the project as an Apache-2.0 open-source OpenAI- and Anthropic-compatible inference server for Apple Silicon, with the repository showing more than 4,000 stars on the Rapid-MLX website.

    2026-10-12

    The esp32-thinks repository identified version 0.1.0 as the initial spoken-replies release for Muse Gadget boards; subsequent 0.2.x work added a configurable multi-model brain, wake word, language behavior and an optional openmuse agent.

    2026-07-24

    Released Rapid-MLX 0.11.0, adding a built-in MCP-capable chat agent, five model families, schema-valid tool calls, prefix/response caching and additional audio/vision capabilities.

    2026-03-05

    Published the first Rapid-MLX benchmark article, evaluating 17 MLX models for coding, reasoning, general knowledge and tool calling on an M3 Ultra; the article was updated October 6, 2026.

    Key Capabilities

    10
    Local inference on Apple Silicon with no account or cloud inference required
    OpenAI-compatible and Anthropic-compatible HTTP APIs
    Reliable schema-valid tool calling for coding-agent workflows
    MCP agent loop and integration with external MCP servers
    Continuous batching, prefix/KV caching, response caching and speculative decoding
    Support for text, vision, audio and video MLX-native models

    Integrations & Partnerships

    Platform Integrations

    • macOS 14+ on Apple Silicon
    • OpenAI-compatible API at localhost:8000/v1
    • Anthropic-compatible API
    • Homebrew, pip/uv and shell-install distribution
    • Claude Code, Cursor, Cline, Continue, Goose, OpenCode, Pi, Aider and Codex CLI
    • ComfyUI workflows for local image and video generation
    • MCP servers
    • ESP32-S3/Muse Gadget and Seeed SenseCAP Watcher W1-B via esp32-thinks

    Key Partnerships

    Built on Apple's MLX framework and MLX model ecosystem
    esp32-thinks is built on Meta's open-source muse-gadget-sdk
    The esp32-thinks openmuse component integrates with OpenAI-compatible upstream services and local endpoints

    Connect

    Website
    rapidmlx.com/
    GitHub
    raullenchai
    X / Twitter
    raullen

    AI Topics

    6

    Raullen Chai focuses on these topics:

    AI Coding Assistants(1)
    Speech Recognition(1)
    Conversational Agents(1)
    Voice Assistant(1)
    LLM Orchestration(1)
    Local Inference(1)
    Back to all developersSuggest an edit