Developing generative AI solutions
Deploying the Application
Putting it into service
The story: Opening a new branch of your shop: you think about the building layout, whether it can handle rush hour, how to keep it secure, and whether customers find it easy to use.
In AI/AWS terms: Deployment integrates the model into its target environment, with attention to architecture, scalability, security, and user experience.
For the exam: Deployment considers architecture, scalability, security, and user experience.
Things to check
The story: Before opening:
- Rent is pay-as-you-go, with no minimum charge.
- Some suppliers only deliver to certain cities.
- The building has a maximum capacity, so check it fits your crowd.
- The landlord secures the building; you lock your own stockroom. And if you send stock to a warehouse outside the building, think about the risk of it leaving your hands.
In AI/AWS terms:
- Cost: pay for what you use, with no minimum fees.
- Regions: some models are only available in certain AWS Regions.
- Quotas: make sure your account has enough service capacity.
- Security: on AWS, security is a shared responsibility between you and AWS. If you access a model outside AWS, evaluate the risk of data leaving your account.
For the exam: Check cost, Region availability, quotas, and security. Security is shared between you and AWS.