What is an AWS Machine Learning PoC?
The first step in a Machine Learning Proof of Concept (PoC) is an assessment of your data. Once's that . This migration is generally advisable to address the following ten steps:
- Gathering business requirements
- Data Assessment
- Data Storage
- Data Transfer
- Data Security
- Data Preparation
- Training & Model Development
- Monitoring & Management
- Cost Optimization
- Documentation & Collaboration
If you are just getting started, you may want to learn more about migration of your data to the cloud.
How much money does an AWS ML PoC cost?
The hard costs can range from $5k to $25k for AWS Machine Learning Proof of Concepts, however these costs can be offset or completely covered by AWS depending on your annual AWS spend. In many cases by involving a qualified AWS Consulting Partner they can be absolutely free for the end customer!
Going it alone on machine learning can be a very expensive proposition. There are several layers of processing: prep, training, production, and ongoing monitoring. Many companies end up not fully accounting for all of the costs, especially those that are ongoing.
How long does an AWS machine learning PoC take?
This largely depends on how much how well the business requirements can be mapping, and reduced to a clear ROI. Once the business requirements are clearly articulated and initial model can be completed as quickly as in a few weeks, while the ongoing processing of this data in a continuing cycle of improvement. Once in production, the model still requires the following:
- Monitoring output
- Checking data quality
- Model performance evaluation
- Gathering feedback from users, stakeholders, and domain experts.
- Model retraining
- Deployment & versioning
- Improving scalability and efficiency.
- Ensuring security & Privacy
What services can an AWS machine learning PoC use?
An AWS machine learning PoC can be built using any of the over 30 AWS machine learning services available. In many cases, the proof of concept will involve multiple services, and be built around a core service or two. Some examples of machine learning PoC services include:
- Amazon SageMaker - Amazon's core build-your-own machine learning framework
- Amazon Rekognition - machine learning computer vision to analyze image and video
- Amazon Comprehend - discover insights and relationships in text
For a complete list of AWS machine learning services, along with an overview of categories (Deep Learning, Add-on Services, etc.) check out our AWS machine learning page.
Getting started
If you are interest in learning more about getting started with your company's own machine learning proof of concept, contact us today our purchase it directly in the AWS Marketplace!