# Humanlaya

- **Website:** <https://humanlaya.com>
- **Primary alias:** `humanlaya.com`

## Description

Humanlaya is an innovative startup dedicated to advancing artificial general intelligence (AGI) through expert-level data provision. With a team comprised of seasoned professionals from leading AI labs, Humanlaya specializes in delivering high-quality data solutions tailored for large language models. Their offerings encompass a wide array of post-training data types including Reinforcement Learning with Human Feedback (RLHF), Supervised Fine-Tuning (SFT), and evaluation rubrics. Humanlaya positions itself as a strategic data partner for top AI laboratories, combining expert networks and a robust quality control system to ensure peak performance in model training. The company is bolstered by investments from notable backers such as Hongshan China and BAI Capital, further establishing its credibility in the AI sector. Whether it’s instructional data for STEM-related queries or services like expert professional domains and code writing, Humanlaya is dedicated to bridging human intelligence with cutting-edge AGI technology, paving the way for future advancements in the field.

## Industries

- 🖥 Computers Electronics and Technology
- 🖥 Artificial Intelligence and Machine Learning _(under Computers Electronics and Technology)_

## Links

- [twitter](https://twitter.com/Humanlayadata)

## Logos & icons

- logo _(primary)_ — PNG — [download](https://cdn.brandfetch.io/idzH9cMx_g/w/272/h/272/theme/dark/logo.png?c=1bxid64Mup7aczewSAYMX&t=1778257653566)
- icon _(primary)_ — JPEG — [download](https://cdn.brandfetch.io/idzH9cMx_g/w/400/h/400/theme/dark/icon.jpeg?c=1bxid64Mup7aczewSAYMX&t=1778257653703)

## Colors

| Name | Hex | Theme |
| --- | --- | --- |
| Venice Blue | #045895 | accent |
| Black Pearl | #020617 | dark |
| White | #ffffff | light |

## Fonts

- Inter — asset _(google)_

---

_Source: <https://brandfetch.com/humanlaya.com>_
_See [/llms.txt](/llms.txt) for a full list of agent-readable pages, and [/auth.md](/auth.md) for how agents authenticate._