When I first started learning AI, most of my attention naturally went to the model itself.
I used to think mainly about questions like:
Which algorithm should I use?
How do I improve accuracy?
How do I preprocess the data?
How well does the model predict?
Those questions still matter.
But over time, as I started building more complete applications, I realized something important:
A good model is only one part of a useful AI system.
Recently, I worked through three FastAPI modules:
CORS
SQL (Relational) Databases
Bigger Applications
At first, these topics may look more like backend engineering than AI.
But that is exactly the point.
The model is not the whole product
A machine learning model in a notebook is not the same thing as a real system.
Once you want real users, APIs, storage, structure, and integration, the problem changes.
Instead of only thinking about predictions, you start thinking about:
how data enters the system
how the frontend talks to the backend
where application data is stored
how the backend is organized
how the project remains maintainable as it grows
That is where FastAPI has started becoming very meaningful for me.
What CORS taught me
CORS may seem like a small technical detail, but it represents a real production issue.
A frontend and backend often run on different origins, and the browser enforces rules around how they communicate.
That means a working model alone is not enough.
The surrounding system also has to be configured properly.
This taught me that deployment and integration matter just as much as model logic.
What SQL databases changed for me
Relational databases were another important step.
In simple demos, it is easy to ignore persistence.
But real applications need to store things such as:
users
requests
results
metadata
application state
logs or records
Working with SQL in FastAPI made me think more seriously about how AI applications behave as actual software systems.
A model may produce intelligence, but the application still needs proper data management.
Why bigger applications matter
The “Bigger Applications” module was especially useful because it focused on structure.
Small apps can survive in one file.
Larger ones should not.
As a project grows, clean separation becomes more important:
routers
dependencies
modules
reusable components
maintainable organization
This is one of the clearest reminders that AI engineering is not only about models.
It is also about building software that remains understandable and usable over time.
How my mindset is changing
I still care deeply about Machine Learning and Computer Vision.
That remains the center of my direction.
But I increasingly want to think beyond:
dataset → model → accuracy
and more in terms of:
problem → data → model → API → database → application → real use
That broader view feels much closer to the kind of engineer I want to become.
Not someone who can only train a model.
But someone who can help turn AI into a complete and usable system.
Final thought
I’m still learning, and I’m still early in the process.
But this shift in thinking already feels important.
FastAPI is helping me understand that building AI in the real world means combining:
machine learning
backend engineering
APIs
databases
structure
integration
maintainability
And honestly, that makes the work even more interesting.
Top comments (0)