Resuming tests on Artificial Intelligence in the clouds with the highest consumption such as AWS, Azure, GCP, Oracle, Huawei; now I'm ready to share what I tested in a simple way with AWS, validating what is mentioned for each tool.
After check the documentation:
It was interesting the pilot test thru Sagemaker, Bedrock and Q. The validation to assambly scalable data lakes, datawarehouse, streaming, graph data; adding unified data acces, unified governance and validating cost vs outcomes.
Then, the next step was the pilot to play between AWS IA options thru your tools:
- Choose best models
- Adjust inference parameters
- Incorporate domain-specific knowledge
- Use safeguards to filter an detect contents
- Use human in the loop revews
- Modify prompts
- Add filters
- Integrate auditing
- Autmate process
- Bedrock knowledge bases
- Define disallowed topics
- Block, encryption, access controlled and protect sensitive inormations, detect hallucinations
- Granular control
-Add monitoring, governance, lineage, security features and protect data across te lifecycle
In the following articles it will be the turn of the tools of the other available clouds, the objective at the end of the exercise is to test each option considering that each client has their preferences.
Customers need solutions, which is why you don't sell brands, we create solutions.
By Jorge Mercado
JMCoach
Comentarios
Publicar un comentario