Amazon SageMaker

Amazon SageMaker constricts the use of their own mxnet package and does not offer a strong Kubernetes backbone. At the same time, Kubeflow is still quite buggy and cumbersome to use. Which tool is a better pick for MLOps pipelines (both from the perspective of scalability and depth)?

We have to process the video stream to identify emotions. For which we need to use Amazon Rekognition/custom model on Amazon SageMaker. With Kinesis WebRTC Javascript SDK, currently, video can be streamed only into the kinesis signaling channel. Signaling channel data is available for streaming only and not processing (ML). So, how can we get real-time data for processing into Kinesis Streams from the frontend?

For streaming the video from frontend to backend into the Amazon Kinesis Video Streams for processing, we tested with Kinesis webRTC JavaScript SDK, and we are facing issues while implementing as mentioned above, so would Chime SDK serve as an alternative to this?

In Rekognition, "create-stream-processor" has a settings parameter. This currently only supports FaceSearch. We are looking to Detect and analyze faces. Is that possible with "create-stream-processor" in the Python SDK? Or do we have to use the Java SDK?

For our Compute services, we decided to use AWS Lambda as it is perfect for quick executions (perfect for a bot), is serverless, and is required by Amazon Lex, which we will use as the framework for our bot. We chose Amazon Lex as it integrates well with other #AWS services and uses the same technology as Alexa. This will give customers the ability to purchase licenses through their Alexa device. We chose Amazon DynamoDB to store customer information as it is a noSQL database, has high performance, and highly available. If we decide to train our own models for license recommendation we will either use Amazon SageMaker or Amazon EC2 with AWS Elastic Load Balancing (ELB) and AWS ASG as they are ideal for model training and inference.