Review on Clarifai by Shane Jenkins

- Ability to train models from scratch.- Easy API access (no need of coding).- Highly scalable architecture.- Support both ROCm and Python APIs.- Good support team. I am using it currently mainly because of its scalability but overall experience so far great! No dislikes yet though we are very new user still. Currently working around image recognition problem where i have quite large dataset having about 4GB worth of training samples which needs to get processed quickly without any loss in accuracy or latency. We are solving this issue primarily through Clarify framework along with AWS lambda function for processing of input pipeline. This works really well when trained model's performance does not vary much between different environments. Otherwise you may want multiple instances/clusters running parallelized tasks instead if just one instance can handle all the load otherwise it will become extremely slow once one becomes busy again due to network traffic spikes etc.