README.md

# Appen AI Research

**The data behind the models you use every day.**

Appen sits at the intersection of human expertise and AI development — we've helped build training data for the world's leading search engines, ranking models, and frontier LLMs. Now we're publishing the research, benchmarks, and insights that come from working at that scale.

## What we publish here

- 📄 **Research** — data-centric AI, model performance, model localization and model safety
- 📊 **Benchmarks** — evaluation suites covering hallucination, bias, annotation quality, and model readiness
- 📘 **Whitepapers** — practical guides on responsible AI, data strategy, and the annual State of AI & ML report

## Where our expertise runs deep

- **Alignment & RLHF** — CoT reasoning traces, SME-led feedback, SFT demonstrations, adversarial red teaming
- **Multimodal** — VLM training data, image-text pairs, video annotation, structured document labelling
- **Speech & audio** — 500+ global locales, expressive TTS, dialectal and code-switched speech
- **Evaluation** — hallucination benchmarking, A/B arena testing, bias detection, regulatory audit support
- **Embodied AI** — World Model Data, Egocentric Data, Robotics, LiDAR.

## Learn more

Interested in our Research, our Benchmarks or how we can support your GenAI efforts? Reach out to [researchteam@appen.com](mailto:researchteam@appen.com)

🌐 [appen.com](https://www.appen.com)  |  📘 [Whitepapers](https://www.appen.com/whitepapers)  |  🧪 [Data-Centric AI](https://www.appen.com/data-centric-ai)