I have 10+ years experience with Amazon AWS & have worked with new features of Amazon as they have added them like DynamoDB, Lambdas, S3 buckets etc.
I also have 15+ years experience with Python & have worked on many projects with node.js, opencv & other similar facial recognition libraries like Tensorflow, dlib.
However Amazon simplifies our life here as they recently launched "Amazon Rekognition" which takes out the requirement of having dedicated GPU cards, huge library dependencies & heavy CPU's just for the purpose of facial recognition.
I can write a service that runs in Lambdas of Python, Golang or node.js(choose which you are comfortable with) & hands over the heavy lifting of FR to Amazon Rekognition.
As discussed I'm reducing the documentation and revising my offer