潜力新星（90天）

997 playable, source-attributed MiniMax H3 / Hailuo H3 video cases, 309 complete public Prompts, and 13 source-checked guides.

MLOps — vLLM + LiteLLM + Portainer CE + Netdata. Case study song ngữ: kiến trúc, metrics, bài học vận hành. · Self-hosted LLM inference done the MLOps way, a bilingual case study

📋 Clipboard manager for vibe coders — native macOS menu-bar app. Text & image history, screen capture + annotation, scrolling capture, screen recording → video/GIF, OCR, on-device voice & meeting notes (or OpenAI/Gemini), share links from your own S3 bucket. Free, private, open source.

Making MoE fine-tuning accessible to anyone with consumer hardware.

CPU-first, eager-execution tensor/autograd runtime and LLM inference engine written in pure Zig

Local-first AI coding agent runtime — your models, fail-closed safety, evidence-gated self-improvement

本地优先的 macOS 与 Windows 语音输入工具，支持端侧 SenseVoice 识别、全局语音输入、文件转写及可选 AI 校对。Local-first voice dictation for macOS and Windows with on-device SenseVoice, typing into any app, file transcription, and optional AI proofreading.

Hands-on GPU/HPC infrastructure operations: K8s GPU scheduling, HAMi sharing, Slurm, observability & vLLM inference. Learn it free on a laptop; validate on one cheap GPU.

The complete open-source LLM curriculum — prompt engineering, RAG, fine-tuning, agents, and production deployment. Beginner to advanced. Python. Free forever.

KoutenDB: locality-aware NoSQL document/vector database built around rings and orbit-inspired retrieval

The in-browser vector store that remembers — a Rust→WASM HNSW engine that persists to OPFS and stays consistent across tabs. Private, offline semantic search in 3 lines.

Local LLM speed measurements from a 4x RTX 3090 rig, with exact launch commands and per-run provenance.

A low-token visual evidence compiler for text-only coding agents. Convert images into compact Visual Evidence Packets (VEP) for DeepSeek, Codex, Claude Code, and other text-only models.

Codex Skill：让纯文本模型（DeepSeek）借助 StepFun step-3.7-flash 获得看图能力 | Give text-only Codex models (DeepSeek) image understanding via StepFun step-3.7-flash

Give non-multimodal Claude Code main models the ability to see pasted screenshots — a ~200-line UserPromptSubmit hook.

CLI & async Python library for free AI chat, image & video generation.

Thinking-native compiler for LEET Language Understanding item sets — killer-item design grammar, verification separation, deterministic gate chain, 3-model benchmark

Hikari's curated English measured notes on local LLM systems (serving, specdec, training, steering).

Your AI/ML research, on one deck — aggregate arXiv/HF/Semantic Scholar papers, star them, and read in-app with cross-device scroll resume.

Reproducible recipe: serve abliterated Gemma-4-12B (gemma4_unified) at 50-118 tok/s on no-NVLink Blackwell (SM120) via vLLM nightly + ModelOpt FP8/NVFP4 + MTP spec-decode.