Iris: AI Model Training Skills

WARNING

🚧 Iris is not yet released.

This module is under active development and will be available in a future Telekinesis SDK release.

INFO

Iris is a module in the Telekinesis SDK for training, fine-tuning, and deploying state-of-the-art deep learning models.

It provides a flexible framework for managing datasets, training pipelines, and model deployment, enabling developers to integrate advanced AI capabilities into robotics and Physical AI applications.

When to Use Iris?

Use IRis for robotics applications that require custom deep learning models or fine-tuning existing models, such as:

• Training perception models for vision-guided pick-and-place
• Fine-tuning object detection or segmentation models for robot-specific datasets
• Deploying models for real-time inference on robots
• Customizing foundation models or VLMs for robotic reasoning tasks
• Integrating learned models into Physical AI agent pipelines

What Does Iris Provide?

IRis includes a collection of modular skills for:

• Dataset management and preprocessing
• Model training and fine-tuning pipelines
• Evaluation and performance metrics
• Model deployment for real-time or batch inference
• Integration with other Telekinesis modules (Vitreous, Retina, Cornea, Pupil)