Amelue Technologies Inc.

We solve hard problems in AI. Our particular expertise is in bringing structure to unstructured, unlabeled, and unorganized data. 

Above: 280k Organ Embeddings from a Self-Supervised Vision Transformer for 140 organs from 3D CT and MRI Images. 

We specialize in giving structure to both 2D and 3D unlabelled images. This is perfect for image similarity search. 

See what the AI sees. AI often works like a black box,  but it doesn't have to be this way. We believe that you should literally see what the AI sees. And, we believe, when an AI provides inference, it should also provide supporting evidence for it. 

We offer off the shelf open-source solutions for both 2D and 3D image segmentation. We also offer custom Vision Transformer-based solutions made from scratch to meet your exact needs. 

We can make your Large Language Model run fast. We re-implement open-source solutions like Llama to use Xformers fast attention and PyTorch's Fully Sharded Data Parallel or Distributed Data Parallel for fast inference and fine-tuning.

We can make custom solutions to extract many useful features from images and text.