Does this skill support data versioning?

Yes, it provides implementation guidance for LakeFS, allowing developers to use Git-like operations (branch, commit, merge) for their AI data lakes.

Which orchestration tools are supported for data workflows?

It includes patterns for modern orchestration frameworks including Dagster, Prefect, and Airflow to manage complex AI data transformations and asset lineage.

What embedding models does this skill recommend?

The skill recommends Voyage AI (voyage-3) for maximum retrieval quality and OpenAI's text-embedding-3-small for cost-sensitive development and prototyping.

How does it help improve RAG quality?

It provides standardized 512-token chunking strategies and integrates RAGAS evaluation metrics to measure factual consistency and answer relevancy.

AI Data Engineering

Name: AI Data Engineering
Author: ancoleman

byancoleman

•

158

データサイエンスとML

Builds production-ready data infrastructure for AI systems, including RAG pipelines, feature stores, and embedding workflows.

概要

The AI Data Engineering skill equips Claude with specialized capabilities to architect and implement robust data pipelines for modern AI and Machine Learning applications. It provides detailed implementation patterns for Retrieval-Augmented Generation (RAG), including ingestion, chunking, and semantic search, alongside advanced data management strategies like feature stores and versioning. By leveraging best-in-class tools such as Voyage AI, Feast, and Dagster, this skill ensures that developers can build scalable, high-quality data backends that power sophisticated AI chat interfaces and search systems.

主な機能

Automated RAG evaluation using RAGAS metrics for faithfulness and relevancy
Advanced workflow orchestration patterns for Dagster, Prefect, and Airflow
Feature store integration with Feast and Tecton for real-time ML serving
158 GitHub stars
End-to-end RAG pipeline architecture covering ingestion, indexing, and retrieval
Optimized chunking strategies and embedding generation using Voyage AI and OpenAI

ユースケース

Developing scalable ML feature pipelines to eliminate training-serving skew
Building high-performance semantic search engines with vector databases like Qdrant
Implementing enterprise RAG systems with automated evaluation and monitoring

概要

主な機能

Automated RAG evaluation using RAGAS metrics for faithfulness and relevancy
Advanced workflow orchestration patterns for Dagster, Prefect, and Airflow
Feature store integration with Feast and Tecton for real-time ML serving
158 GitHub stars
End-to-end RAG pipeline architecture covering ingestion, indexing, and retrieval
Optimized chunking strategies and embedding generation using Voyage AI and OpenAI

ユースケース

Developing scalable ML feature pipelines to eliminate training-serving skew
Building high-performance semantic search engines with vector databases like Qdrant
Implementing enterprise RAG systems with automated evaluation and monitoring