teaser.mp4

AIE-Skills

Agent Skills cho AI/ML Engineering — từ setup Python project, fine-tune LLM, đến deploy inference server.

---

Install

npx skills add jayll1303/AIEKit

Install specific skills:

npx skills add jayll1303/AIEKit --skill ultralytics-yolo --skill vllm-tgi-inference

Install globally (available across all projects):

npx skills add jayll1303/AIEKit -g

List available skills:

npx skills add jayll1303/AIEKit --list

Requires Node.js >= 18. Uses skills.sh ecosystem.

---

Skills (34)

Skill	Mô tả
aie-skills-installer	Analyze target project codebase và đề xuất chỉ cài skills cần thiết
arxiv-reader	Đọc và phân tích paper arXiv qua HTML
disk-cleanup	Diagnose and clean disk space on Linux/ML servers with Docker
docker-gpu-setup	Dockerfile & docker-compose cho GPU/CUDA workloads
experiment-tracking	Selfhosted experiment tracking với MLflow / W&B
fastapi-at-scale	Production-grade FastAPI: async SQLAlchemy, Alembic, JWT, rate limiting, testing
freqtrade	Crypto trading strategies với Freqtrade
hf-hub-datasets	Download, upload, stream models & datasets từ HuggingFace Hub
hf-speech-to-speech-pipeline	Queue-chained speech pipeline: STT/LLM/TTS handlers, VAD, streaming
hf-transformers-trainer	Fine-tune & align LLMs với Trainer, TRL, PEFT (LoRA/QLoRA)
k2-training-pipeline	Train speech models với Next-gen Kaldi: k2, icefall, lhotse
llama-cpp-inference	Chạy GGUF models locally với llama-server, llama-cli, llama-cpp-python
ml-brainstorm	Brainstorm ML/AI decisions: training, model selection, serving, quantization
modal-batch-processing	Modal job orchestration: .map, .starmap, .spawn, .spawn_map, @modal.batched
modal-sandbox	Modal Sandbox lifecycle: isolated execution, tunnels, snapshots, file IO
model-quantization	Quantize LLMs với GGUF, GPTQ, AWQ, bitsandbytes
notebook-workflows	Tạo & chỉnh sửa Jupyter/Colab notebooks programmatically
ollama-local-llm	Chạy local LLMs với Ollama: pull, run, Modelfile, REST API
openai-audio-api	OpenAI-compatible audio/speech APIs với FastAPI, dynamic batching
opentelemetry	Distributed tracing, metrics, logs với OpenTelemetry
paddleocr	OCR với PaddlePaddle: detection, recognition, fine-tuning, PP-OCRv5
python-ml-deps	Cài ML deps với uv, xử lý CUDA version conflicts
python-project-setup	Bootstrap Python projects với uv, ruff, pytest
python-quality-testing	Type annotations, Hypothesis testing, mutation testing
semantic-router	Superfast AI decision layers: Route, SemanticRouter, HybridRouter
sglang-serving	Serve LLMs với SGLang: RadixAttention, structured output
sherpa-onnx	Offline speech: ASR, TTS, VAD, speaker diarization
tensorrt-llm	Optimize LLM inference với TensorRT-LLM: FP8/INT4, kernel fusion
text-embeddings-inference	Deploy embedding/reranker models với HuggingFace TEI
text-embeddings-rag	RAG pipelines với sentence-transformers, FAISS, ChromaDB, Qdrant
triton-deployment	Deploy models trên NVIDIA Triton Inference Server
ultralytics-yolo	Train, predict, export YOLO models (detect, segment, pose, OBB)
unsloth-training	Fine-tune LLMs 2x faster, 70% less VRAM: SFT/DPO/GRPO
vllm-tgi-inference	Serve LLMs locally với vLLM hoặc TGI

---

Profiles

Skills được nhóm theo domain:

Profile	Skills	Mô tả
Core	aie-skills-installer, python-project-setup, python-ml-deps, hf-hub-datasets, docker-gpu-setup, notebook-workflows	Foundation cho mọi AI/ML project
LLM	hf-transformers-trainer, unsloth-training, model-quantization, experiment-tracking	Fine-tune LLMs
Inference	vllm-tgi-inference, sglang-serving, llama-cpp-inference, ollama-local-llm, tensorrt-llm, triton-deployment	Deploy LLM servers
Speech	k2-training-pipeline, sherpa-onnx, hf-speech-to-speech-pipeline, openai-audio-api	Speech processing
CV	ultralytics-yolo, paddleocr	Computer vision
RAG	text-embeddings-rag, text-embeddings-inference, semantic-router	RAG pipelines
Backend	fastapi-at-scale, opentelemetry, python-quality-testing	API & observability
Modal	modal-batch-processing, modal-sandbox	Modal platform orchestration

Standalone: arxiv-reader, disk-cleanup, freqtrade, ml-brainstorm

---

Smart Install

Sau khi cài, dùng skill aie-skills-installer trong agent — nó sẽ:

1. Scan codebase (deps, imports, Dockerfiles, notebooks...)
2. Recommend chỉ skills phù hợp với project
3. Chờ confirm trước khi cài thêm

---

License

MIT