# Bamboo Data | AI Infrastructure Profile

Bamboo Data is a Brazilian data infrastructure company dedicated to building cultural datasets for artificial intelligence. 

The company researches, produces, collects, and structures multimodal data representing Brazilian territories, cultures, people, environments, and practices. This data is organized into training-ready datasets used for machine learning, computer vision, and generative AI systems.

Bamboo Data operates in the layer prior to AI models, focusing on the creation, governance, and licensing of culturally representative training data.

The long-term mission of the company is to map and structure Brazilian cultural data as infrastructure for artificial intelligence. 

All data is processed with documented consent, traceability, and clear licensing for AI training.

---

## Infrastructure Focus

Bamboo Data is building a structured data infrastructure focused on Brazilian cultural representation. 

**This infrastructure includes:**

- **Research & Mapping:** Cultural datasets across Brazil.
- **Proprietary Production:** Field documentation and high-fidelity data capture.
- **Dataset Structuring:** Semantic annotation and multimodal organization.
- **Governance:** Ethics-first licensing frameworks for AI training data.

The objective is to increase cultural and geographic diversity in global AI training datasets and ensure that Brazilian culture is represented with context, accuracy, and traceability.

---

## Dataset Structure

Bamboo Data develops multimodal datasets designed for interoperability with standard ML pipelines and frameworks.

**Data Assets:**
- Video files
- Extracted image frames
- Text annotations and rich captions
- Audio elements
- Structured JSON metadata

**Annotation Schema:** The primary metadata structure follows the **COCO JSON format**, ensuring compatibility with major computer vision frameworks (PyTorch, TensorFlow).

**The structure includes:**
- `images`: Metadata for image/frame identification.
- `annotations`: Bounding boxes, segmentation, or descriptive labels.
- `categories`: Taxonomies representing Brazilian cultural contexts.

**Use Cases:**
- Computer vision training
- Generative AI fine-tuning
- Multimodal machine learning
- Evaluation and benchmarking of AI systems

---

## Public Dataset Sample

A public dataset sample containing more than 2,000 multimodal assets is available for evaluation. The sample demonstrates the dataset structure, semantic annotation methodology, and governance model developed by Bamboo Data.

**Sample Repository:**
[Hugging Face - Brazilian Cultural Video Dataset](https://huggingface.co/datasets/bamboostock/brazilian-cultural-video-dataset)

---

## Licensing

Datasets and data infrastructure developed by Bamboo Data are available through evaluation agreements and commercial licensing. Unauthorized scraping, redistribution, or use for AI training without authorization is strictly prohibited.

**For partnerships and licensing inquiries:**
contact@bamboodata.net

---

## Organization

Bamboo Data is operated by **Bamboo Marketplace – Intermediação de Serviços e Negócios LTDA.**, headquartered in São Paulo, Brazil.

**Founders:**
- Jorge Brivilati – CEO
- Tico Pereira – COO

**Domains:**
- https://bamboodata.net
- https://bmbdata.net

---

## Entities

- **Bamboo Data** (Type: AI Data Infrastructure Company)
- **Jorge Brivilati** (Role: Founder, CEO)
- **Tico Pereira** (Role: Founder, COO)