6. Speech Vision Language
AI PLATFORM ELEMENTS
Machine
Learning
Tools
APIs for Artificial Intelligence
Machine Learning Libraries
Cloud infrastructure
Notebooks
AI Pipelines
Keras Pytorch TensorFlow Scikit-learn
GPU Data storage IoT
11. KEY FACTORS
1. Cloud platform choice
2. Data ownership and privacy
3. Language
4. API availability
5. Cost
HOW TO CHOOSE AI PLATFORM
12. CLOUD PLATFORM CHOICE
Select most of
the components
from one vendor
PREFERRED
CLOUD
PLATFORM
Use the ”best-
of-breed”
approach
HYBRID
SOLUTIONVS.
13. Who owns the data?
DATA OWNERSHIP AND PRIVACY
Does the
API provide
pretrained
data?
PRETRAINED
DATA
Who owns the
training data?
DATA
OWNERSHIP
Where is the
training data and
the predicted
data stored?
DATA LOCATION
14. 14
Entäpä suomen kieli?
Eller fungerar det på svenska?
En werkt het ook in het Nederlands?
Translation to English:
Does this function in Finnish?
How about Swedish?
And is all this possible also in Dutch?
18. What is the charging basis for APIs?
Subscription-based
Transaction-based
CPU-based
Or something else?
COST
19. TAKE-AWAYS
Decide whether to use hybrid or preferred cloud
Who owns the data and where is it stored
Remember the language (limitations)
Check the API availability in your region
Estimate the life-cycle cost
TIPS FOR CHOOSING THE ULTIMATE AI PLATFORM
APIs for AI: For developers
ML Libraries: For data scientists
Cloud infrastructure: For MLOPS
SAP Leonardo AI uses Google AI underneath.
Open-source libraries are always available as well.
Choose all the components from one cloud provider or use hybrid solution
Pretrained data is important when dealing with large data quantities (images, videos, text)
Language recognition works for most languages. The more detailed NLP APIs work for major languages only.
There are workarounds: free text content can be translated into English and the NLP functions be applied to the English content.