Cloud Benchmarking Case Study

Business Problem:

Client wanted consultation on most appropriate Cloud model i.e. Public vs Hybrid as part of their AI adoption strategy for AI scenario development at scale. Expected outcome was Cloud benchmarking across Microsoft Azure, AWS and GCP for image classification and text analytics scenarios .

Proposed Solution:

Cognilytic solution involved model performance testing across the three platforms-

Amazon Web Services APIs- AWS Rekognition and AWS Comprehend.

Google Cloud Platform APIs- Cloud Vision and Natural Language.

Azure ML Package for Text Analytics (AMLPTA) and Azure ML Package for Computer Vision (AMLPCV).

Fully managed platform modeling for AWS Sagemaker and GCP ML Engine using Tensorflow and MXNet frameworks.

Key Deliverables:

Benchmarking results across Azure AWS and GCP on following metrics-

Model Training Time

Model Validation Accuracy

Ease of Adoption