Develop Prototype and Proof-of-Concept (POC)
At DreamAI, we develop and build AI-based applications for corporate customers and startup companies to help develop business case and secure funding. Customization and adaptation of common Deep Learning models and architectures for business use cases.
Build Deep Learning Models for Medical Images
Building State-of-the-art deep learning models for medical image analysis. Developing Machine Learning applications with complete software life-cycle with Active Learning, iterative supervised learning, querying users and other data sources.
AI Infrastructure Management
Infrastructure management for AI applications with full DevOps life cycle including continuous integration, deployment and automated upgrades. AWS Cloud based ML project deployment using SageMaker
Customized Model Building and Evaluation
We would normally train a model on our own local GPUs or in the Cloud up to a reasonable limit without any additional cost. For large scale training requirements that may warrant multiple GPUs to save training time or because of the sheer size of data, we would include the cost of Cloud based GPU training in the project budget. We can use both Pytorch and Tensorflow for training and evaluation, but we can provide the trained model in a portable format like ONNX for deployment and integration
Image Labeling
A critical part of a Computer Vision project that employs Deep Learning algorithms is Image Labeling. We provide services for different types of labeling including but not limited to class tags, bounded boxes, semantic segmentation, instance segmentation.