🌏 China | 🤖 Building AI companions | 🚀 Full-stack agentic engineer
Building the future of human-AI interaction — from terminal agents to desktop companions with voice, vision, and presence.
- 🤖 Ani — AI-powered desktop companion with immersive scenes, real-time video interaction, and natural voice conversation
- 🧠 Codev — Collaborative Agentic Workspace combining search, browser control, coding, voice, and long-session continuity
- 🧠 Omni — Multimodal LLM training & inference framework (LM, VLM, speech) from scratch, with real-time voice/video call
- 📚 CS336 — Hands-on implementation of Stanford CS336: LLM training, tokenization, transformers, and scaling
- 🔄 Duplex — Real-time bidirectional translation with voice I/O and unified UI
- 🤖 Ani — AI-powered desktop companion with wallpaper integration, video layers, companion mode, and plugin system
- 🎙️ Voice Engine — STT/TTS pipeline with VAD, noise suppression, and multi-provider support (Groq, faster-whisper, edge-tts)
- 🧩 Plugin Runtime — Modular plugin system with card protocol, UI panels, and SDK discovery
- 🧠 Codev — Full-featured agentic coding workspace with ToolSearch, SubAgent Swarm, and auto-compact context management
- 🖼️ ImageShowTool — Terminal-native image rendering with Kitty protocol, ink-picture, and aspect-ratio-aware display
- 🔍 WebSearch + WebFetch — Multi-backend search (SearXNG, Tavily, Jina) with image search and terminal image results
- 🗺️ LocationTool — Geographic search with Amap & Google Maps, unlimited paginated results
- 🔌 Bridge System — CCR remote forwarding, AskUserQuestion decoupling, DingTalk & Feishu integration
- 🧠 Omni — Multimodal LLM framework: train LM, VLM, and full speech models from scratch; real-time voice/video call
- 🔄 Duplex — Real-time bidirectional translation with voice I/O, Groq Whisper ASR, and unified Qt UI
- 📚 CS336 — Transformer implementations from scratch, GPT-2 training, scaling laws
- 🔧 System Tooling — Custom IPC, auto-update pipelines, data epoch migration, cross-session collaboration
- Building Ani — Creating an AI companion that feels truly present: immersive wallpaper scenes, real-time voice, and natural interaction
- Expanding Codev — Making agentic development workflows faster, smoother, and more collaborative
- Building Omni — A from-scratch multimodal LLM framework spanning text, vision, and speech
- Studying LLMs — Working through Stanford CS336 to deepen understanding of transformer internals
- Rapid prototyping — Full features in days, iterating on real user feedback
"Build what you'd want to use yourself" — Every project starts as a personal need, then grows into something others can benefit from.
Random Facts
- Ani started as a weekend experiment, grew into a full desktop companion platform
- Runs multiple AI models in parallel for different tasks
- Deeply invested in making AI interaction feel natural, not robotic
- Constantly exploring new ways to bridge terminal AI with rich desktop experiences