AI is a scoring machine to automate data processing to get an inference.
Quality of the data defines the quality of the AI
Quality of feature engineering either does
No domain knowledge - no AI
Data Scientist works on data and models, to build an application Engineers and DevOps are required
In the real world a model will encounter a case it was never trained for. Continuous Retraining is a must.
10. Tasks
10
● Predicting the demand [for coffee, upon the weather forecasts]
● Advertising budgets allocation
● Predicting the malfunction of a drill rod
[Multi]Linear/Logistic Regression
19. Neural Networks: error back propagation
19https://towardsdatascience.com/https-medium-com-piotr-skalski92-deep-dive-into-deep-netwo
rks-math-17660bc376ba
Y = WX + B
X → → Y
29. AI and ML team
29
AI = Task + [ Data ] + Algorithm
Domain
Expert
Data
Scientist
Application
[Production]
Data
Predictions to the
consequent
operations
DevOps
+
Software Engineer
Runtime Infrastructure
41. Takeaways
41
● AI is a scoring machine to automate data processing to get an
inference.
● Quality of the data defines the quality of the AI
● Quality of feature engineering either does
● No domain knowledge - no AI
● Data Scientist works on data and models, to build an application
Engineers and DevOps are required
● In the real world a model will encounter a case it was never trained
for. Continuous Retraining is a must.