orchestra-research-ai-research-skills

Overview

AI Research Engineering Skills Library by Orchestra Research is the most comprehensive open-source collection of AI research engineering skills for AI coding agents. With 83 skills across 20 categories, it provides the engineering layer that enables coding agents to write and conduct AI research experiments, from dataset preparation and training pipelines to model deployment and agent building.

Key Features

• 83 Skills Across 20 Categories — Covers model architecture, fine-tuning, post-training, distributed training, optimization, inference, RAG, multimodal, safety & alignment, MLOps, mechanistic interpretability, and more
• Research-to-Code Pipeline — Bridges the gap between AI research hypotheses and executable experiment infrastructure
• Multi-Agent Compatible — Works with Claude Code, Codex, and other AI agents via npm package or direct installation
• Structured Skill Format — Each skill includes context, methodology, best practices, and agent-ready instructions for consistent quality
• Active Community — Slack community and regular updates with new research categories

Who is this for?

AI researchers and ML engineers who spend more time debugging infrastructure than testing hypotheses. This library enables coding agents to handle the engineering work — setting up training pipelines, configuring distributed training, deploying models — so researchers can focus on the science.