​
Get Started

OcularAI SDK is a developer tool designed to streamline dataset management for vision-language models. The SDK makes it easy to integrate OcularAI datasets directly into your ML workflow with minimal effort.

The OcularAI SDK is currently in its initial release phase with a focus on dataset downloading and management features. We’re actively working on expanding its capabilities.

​
Key Features

​
Initial Release Features

Our initial release focuses on essential functionality to make your workflow smooth and efficient:

  • Simple Dataset Export - Easy-to-use URLs for downloading dataset version exports
  • API Integration - Easily accessible with API keys in your vision language model work environment
  • Reliable Downloads - Automatic retry mechanism for reliable downloads with detailed progress tracking

​
Developer-First Approach

We’ve built the SDK with developers in mind, focusing on a clean, intuitive experience:

  • Intuitive API Design - Simple, consistent API that follows modern Python best practices
  • Comprehensive Error Handling - Built-in retry mechanism and detailed error messages
  • Detailed Logging - Robust logging for debugging and progress updates
  • Flexible Configuration - Environment variable support for easy customization

The OcularAI SDK follows Python best practices and is designed to integrate seamlessly with standard machine learning workflows and pipelines.

​
Architecture

The OcularAI SDK follows a hierarchical structure that mirrors the Ocular platform:

Ocular
  ├── Workspace
  │     ├── Project
  │     │     ├── Version
  │     │     │     ├── Export
  │     │     │     │     └── Dataset
  │     │     │     └── ...
  │     │     └── ...
  │     └── ...
  └── ...

This structure allows you to:

  1. Initialize the SDK with your API key
  2. Access your workspace
  3. Select a project
  4. Choose a specific version
  5. Download an export as a dataset

This hierarchical structure aligns with how you organize your data in the OcularAI Foundry platform, making it intuitive to locate and access your datasets.

​
Use Cases

The OcularAI SDK is particularly useful for:

  • Machine Learning Engineers - Easily integrate vision datasets into training pipelines
  • Data Scientists - Quickly access and explore curated datasets for experimentation
  • Research Teams - Streamline dataset access across multiple team members
  • ML Ops - Incorporate dataset versioning into your ML operations

For teams working with complex data pipelines, the SDK enables reproducible workflows by ensuring everyone on the team has access to the same dataset versions.

​
Coming Soon

Future SDK releases will include:

  • Dataset Transformation - Tools for preprocessing and transforming datasets
  • Dataset Creation - Programmatic dataset creation and modification
  • Streaming API - Stream datasets directly into memory for immediate use
  • Framework Integrations - Direct integration with PyTorch, TensorFlow, and other ML frameworks

Ready to get started? Head over to the Quickstart Guide to begin using the OcularAI SDK in your projects.