Learning Objectives

By the end of this chapter, you will be able to:

Understand why ORMs are used.
Explain what Object Relational Mapping (ORM) is.
Understand the overall SQLAlchemy workflow.
Identify the major SQLAlchemy components.
Explain how Python objects are stored and retrieved from a database.

Topics Covered

In this module, you’ll learn:

Introduction to ORM
Building Your First SQLAlchemy Application
CRUD Operations
Retrieving Results
ORM Relationships

Try Yourself: 💻 VS Code | 🚀 Colab | 📥 Download
Verify Solutions: 💻 VS Code | 🚀 Colab | 📥 Download

Working Without an ORM

A relational database understands only SQL. Whenever an application needs to store, retrieve, update, or delete data, the application must send SQL statements to the database. For example, to retrieve all students:

SELECT * FROM students;

To insert a new student:

INSERT INTO students (name, age)
VALUES ('Rahul', 22);

As applications grow, writing and maintaining SQL for every database operation becomes repetitive and difficult.

Working With an ORM

With an ORM, we work with Python classes and objects instead of writing SQL for most database operations. To insert a new student:

student = Student(name="Rahul", age=22)

session.add(student)
session.commit()

To retrieve all students:

students = session.scalars(select(Student)).all()

Notice that we never wrote an SQL query. SQLAlchemy automatically generates the required SQL, sends it to the database, retrieves the results, and converts them back into Python objects. This allows us to focus on writing Python instead of manually writing SQL.

What is an ORM?

ORM (Object Relational Mapping) is a technique that maps Python objects to database tables. It allows us to work with:

Python	Database
Class	Table
Object	Row
Attribute	Column

Think of an ORM as a translator between Python and a relational database.

Python Objects
      ↓
     ORM
      ↓
SQL Queries
      ↓
Database

Throughout this workshop, we’ll use SQLAlchemy ORM, one of the most popular ORM libraries in Python.

The Big Picture

Before learning the individual components, let’s understand the complete workflow.

Writing Data (INSERT / UPDATE / DELETE)

Create Engine
      ↓
Create Session
      ↓
Create Python Object
      ↓
Session Tracks Changes
      ↓
SQLAlchemy Generates SQL
      ↓
Database Executes SQL
      ↓
Data Stored

Reading Data (SELECT)

Create Engine
      ↓
Create Session
      ↓
Build Python Query
      ↓
SQLAlchemy Generates SQL
      ↓
Database Executes SQL
      ↓
Database Returns Rows
      ↓
SQLAlchemy Maps Rows to Python Objects
      ↓
Python Receives Objects

Notice the important role of SQLAlchemy.

We write Python code.
SQLAlchemy converts it into SQL.
The database executes the SQL.
SQLAlchemy converts the returned rows back into Python objects.

We’ll learn each step of this workflow throughout this module.

Understanding the Components

Now let’s understand the responsibility of each component.

Engine

The Engine is the starting point of every SQLAlchemy application. It knows:

Which database to connect to.
How to establish the connection.
How to send SQL statements.

Application
      ↓
Engine
      ↓
Database

Think of the Engine as the gateway to the database.

Session

The Session is used to interact with the database. It allows us to:

Add new objects
Retrieve objects
Update objects
Delete objects
Commit or rollback transactions

Application
      ↓
Session
      ↓
Engine
      ↓
Database

Think of the Session as your conversation with the database.

ORM Model

An ORM model is a Python class that represents a database table.

class Student(Base):
    ...

Each object created from this class represents one row in the table.

Python Class
      ↓
Python Object
      ↓
Database Row

SQLAlchemy

SQLAlchemy acts as the translator. When we write Python code,

select(Student)

SQLAlchemy generates SQL.

SELECT * FROM students;

When the database returns rows,

id | name | age

SQLAlchemy converts them into Python objects.

Student(id=1, name="Rahul", age=22)

Database

The database stores the data and executes SQL statements. It understands SQL, not Python objects.

Putting Everything Together

                Python Application
                       │
                       ▼
                 Create Engine
                       │
                       ▼
                 Create Session
                       │
            ┌──────────┴──────────┐
            │                     │
            ▼                     ▼
     Create Python Object    Build Python Query
            │                     │
            └──────────┬──────────┘
                       ▼
                 SQLAlchemy ORM
                       │
             Generates SQL
                       │
                       ▼
                   Database
                       │
          Executes SQL & Returns Rows
                       │
                       ▼
                 SQLAlchemy ORM
                       │
        Maps Rows to Python Objects
                       │
                       ▼
                Python Application

Summary

In this chapter, you learned:

Why ORMs are needed.
What Object Relational Mapping (ORM) is.
How SQLAlchemy translates Python code into SQL.
The complete workflow for reading and writing data.
The role of the Engine, Session, ORM Model, SQLAlchemy, and the Database.

In the next chapter, we’ll start building this workflow by creating our first Engine, connecting to a database, defining our first ORM model, and creating our first table.

Building Your First SQLAlchemy ORM Application

Learning Objectives

By the end of this chapter, you will be able to:

Create a SQLAlchemy project.
Connect to a SQLite database.
Define an ORM model.
Create database tables.
Create a Session.
Insert data into the database.

Project Setup

Create a new project.

mkdir sqlalchemy-demo
cd sqlalchemy-demo

Initialize the project.

uv init

Install SQLAlchemy.

uv add sqlalchemy

Project structure

sqlalchemy-demo/
│
├── .venv/
├── pyproject.toml
├── uv.lock
└── main.py

Throughout this chapter, we’ll gradually build the same main.py.

Step 1 — Create the Engine

Every SQLAlchemy application starts by creating an Engine. The Engine knows which database to connect to. Import create_engine.

from sqlalchemy import create_engine

Create a Database URL.

DATABASE_URL = "sqlite:///students.db"

Create the Engine.

engine = create_engine(DATABASE_URL)

Workflow

Application
      │
      ▼
Create Engine
      │
      ▼
Database

At this stage,

Engine is created.
No database operations have happened yet.

Current application

from sqlalchemy import create_engine

DATABASE_URL = "sqlite:///students.db"

engine = create_engine(DATABASE_URL)

Step 2 — Create the Declarative Base

Every ORM model inherits from a common Base class. Import DeclarativeBase.

from sqlalchemy.orm import DeclarativeBase

Create the Base class.

class Base(DeclarativeBase):
    pass

Workflow

Base
 │
 ├── Student
 ├── Employee
 └── Department

SQLAlchemy uses the Base class to keep track of all ORM models. Current application

from sqlalchemy import create_engine
from sqlalchemy.orm import DeclarativeBase

DATABASE_URL = "sqlite:///students.db"

engine = create_engine(DATABASE_URL)


class Base(DeclarativeBase):
    pass

Step 3 — Create an ORM Model

An ORM model represents a database table. Import the required modules.

from sqlalchemy import String
from sqlalchemy.orm import Mapped, mapped_column

Create the model.

class Student(Base):
    __tablename__ = "students"

    id: Mapped[int] = mapped_column(primary_key=True)
    name: Mapped[str] = mapped_column(String(100))
    age: Mapped[int]

Mapping

Python	Database
Class	Table
Object	Row
Attribute	Column

SQL Equivalent

CREATE TABLE students (
    id INTEGER PRIMARY KEY,
    name VARCHAR(100),
    age INTEGER
);

Current application

from sqlalchemy import String, create_engine
from sqlalchemy.orm import DeclarativeBase, Mapped, mapped_column

DATABASE_URL = "sqlite:///students.db"

engine = create_engine(DATABASE_URL)

class Base(DeclarativeBase):
    pass

class Student(Base):
    __tablename__ = "students"

    id: Mapped[int] = mapped_column(primary_key=True)
    name: Mapped[str] = mapped_column(String(100))
    age: Mapped[int]

Step 4 — Create Database Tables

Create all tables.

Base.metadata.create_all(engine)

What happens?

Engine
      │
      ▼
Connect to SQLite
      │
      ▼
students.db exists?
      │
 ┌────┴────┐
 │         │
No        Yes
 │         │
 ▼         ▼
Create     Open
Database   Database
      │
      ▼
Read ORM Models
      │
      ▼
Generate CREATE TABLE SQL
      │
      ▼
Create Tables

Note: For SQLite, the database file is created automatically if it doesn’t already exist. SQLAlchemy then creates the tables. For databases like PostgreSQL or MySQL, the database must already exist before connecting.

Current application

Base.metadata.create_all(engine)

Step 5 — Create a Session

A Session is used to interact with the database. Import Session.

from sqlalchemy.orm import Session

Create the Session.

with Session(engine) as session:
    pass

Workflow

Application
      │
      ▼
Session
      │
      ▼
Engine
      │
      ▼
Database

Think of the Session as your conversation with the database.

Step 6 — Insert a Record

Create a Python object.

student = Student(
    name="Rahul",
    age=22,
)

Add the object to the Session.

session.add(student)

Save the changes.

session.commit()

Refresh the object.

session.refresh(student)

Display the generated ID.

print(student.id)

Workflow

Create Python Object
      │
      ▼
Session.add()
      │
      ▼
Session.commit()
      │
      ▼
SQLAlchemy Generates INSERT
      │
      ▼
Database Stores Row
      │
      ▼
Session.refresh()
      │
      ▼
Updated Python Object

Generated SQL

INSERT INTO students (name, age)
VALUES ('Rahul', 22);

Complete Application

from sqlalchemy import String, create_engine
from sqlalchemy.orm import (
    DeclarativeBase,
    Mapped,
    Session,
    mapped_column,
)


DATABASE_URL = "sqlite:///students.db"

engine = create_engine(DATABASE_URL)


class Base(DeclarativeBase):
    pass


class Student(Base):
    __tablename__ = "students"

    id: Mapped[int] = mapped_column(primary_key=True)
    name: Mapped[str] = mapped_column(String(100))
    age: Mapped[int]


Base.metadata.create_all(engine)


with Session(engine) as session:

    student = Student(
        name="Rahul",
        age=22,
    )

    session.add(student)
    session.commit()
    session.refresh(student)

    print(student.id)

Run the application.

uv run main.py

Output ?

Show Output

Open students.db using the SQLite extension. You should see:

id	name	age
1	Rahul	22

Summary

Congratulations! 🎉 You have built your first SQLAlchemy ORM application. You learned how to:

Create an Engine.
Create a Declarative Base.
Define an ORM Model.
Create database tables.
Create a Session.
Insert data into the database.

CRUD Operations with SQLAlchemy ORM

CRUD stands for:

C – Create (INSERT)
R – Read (SELECT)
U – Update (UPDATE)
D – Delete (DELETE)

These are the four fundamental database operations performed using SQLAlchemy ORM.

Create (INSERT)

SQL

INSERT INTO students (name, age)
VALUES ('Rahul', 22);

SQLAlchemy ORM

session.add(Student(name="Rahul", age=22))
session.commit()

If you need the auto-generated values (such as the primary key), refresh the object.

student = Student(name="Rahul", age=22)

session.add(student)
session.commit()
session.refresh(student)

print(student.id)

Workflow

Create Object
      ↓
session.add()
      ↓
session.commit()
      ↓
Database Row Created

Read (SELECT)

Retrieve All Records

SQL

SELECT * FROM students;

SQLAlchemy ORM

students = session.scalars(select(Student)).all()

Retrieve by Primary Key

SQL

SELECT *
FROM students
WHERE id = 1;

SQLAlchemy ORM

student = session.get(Student, 1)

Retrieve the First Record

SQL

SELECT *
FROM students
LIMIT 1;

SQLAlchemy ORM

student = session.scalars(select(Student)).first()

Workflow

Build Query
      ↓
Execute Query
      ↓
Retrieve Objects

Update (UPDATE)

Step 1 – Retrieve the Object

student = session.get(Student, 1)

Step 2 – Modify the Object

student.age = 25

Step 3 – Save the Changes

session.commit()

Complete Example

SQL

UPDATE students
SET age = 25
WHERE id = 1;

SQLAlchemy ORM

student = session.get(Student, 1)
student.age = 25
session.commit()

Workflow

Retrieve Object
      ↓
Modify Object
      ↓
session.commit()
      ↓
Database Updated

Delete (DELETE)

Step 1 – Retrieve the Object

student = session.get(Student, 1)

Step 2 – Delete the Object

session.delete(student)

Step 3 – Save the Changes

session.commit()

Complete Example

SQL

DELETE FROM students
WHERE id = 1;

SQLAlchemy ORM

student = session.get(Student, 1)
session.delete(student)
session.commit()

Workflow

Retrieve Object
      ↓
session.delete()
      ↓
session.commit()
      ↓
Row Deleted

CRUD Summary

Operation	SQL	SQLAlchemy ORM
Create	`INSERT`	`session.add()` + `session.commit()`
Read	`SELECT`	`session.scalars(select(Student)).all()`
Read by ID	`SELECT ... WHERE id = ?`	`session.get(Student, id)`
Update	`UPDATE`	Modify object + `session.commit()`
Delete	`DELETE`	`session.delete()` + `session.commit()`

Complete CRUD Example

### Create
session.add(Student(name="Rahul", age=22))
session.commit()

###Read
students = session.scalars(select(Student)).all()

### Update
student = session.get(Student, 1)
student.age = 25
session.commit()

### Delete
student = session.get(Student, 1)
session.delete(student)
session.commit()

SQL SELECT vs SQLAlchemy ORM

SQL	SQLAlchemy ORM
`SELECT * FROM students;`	`select(Student)`
`SELECT * FROM students;` (Execute & Fetch All)	`session.scalars(select(Student)).all()`
`SELECT * FROM students LIMIT 1;`	`session.scalars(select(Student)).first()`
`SELECT * FROM students WHERE id = 1;`	`session.get(Student, 1)`
`SELECT * FROM students WHERE age > 20;`	`select(Student).where(Student.age > 20)`
`SELECT * FROM students WHERE age > 20;` (Execute & Fetch All)	`session.scalars(select(Student).where(Student.age > 20)).all()`
`SELECT * FROM students WHERE age > 20 AND city = 'Hyderabad';`	`select(Student).where(Student.age > 20, Student.city == "Hyderabad")`
`SELECT * FROM students WHERE age > 20 OR city = 'Hyderabad';`	`select(Student).where(or_(Student.age > 20, Student.city == "Hyderabad"))`
`SELECT * FROM students ORDER BY name;`	`select(Student).order_by(Student.name)`
`SELECT * FROM students ORDER BY age DESC;`	`select(Student).order_by(Student.age.desc())`
`SELECT * FROM students LIMIT 5;`	`select(Student).limit(5)`
`SELECT * FROM students LIMIT 5 OFFSET 10;`	`select(Student).limit(5).offset(10)`
`SELECT DISTINCT city FROM students;`	`select(Student.city).distinct()`
`SELECT COUNT(*) FROM students;`	`select(func.count(Student.id))`
`SELECT SUM(marks) FROM students;`	`select(func.sum(Student.marks))`
`SELECT AVG(marks) FROM students;`	`select(func.avg(Student.marks))`
`SELECT MIN(age) FROM students;`	`select(func.min(Student.age))`
`SELECT MAX(age) FROM students;`	`select(func.max(Student.age))`
`SELECT city, COUNT(*) FROM students GROUP BY city;`	`select(Student.city, func.count()).group_by(Student.city)`
`SELECT city, COUNT() FROM students GROUP BY city HAVING COUNT() > 5;`	`select(Student.city, func.count()).group_by(Student.city).having(func.count() > 5)`
`SELECT * FROM students WHERE name LIKE 'R%';`	`select(Student).where(Student.name.like("R%"))`
`SELECT * FROM students WHERE age BETWEEN 20 AND 25;`	`select(Student).where(Student.age.between(20, 25))`
`SELECT * FROM students WHERE city IN ('Hyderabad', 'Bangalore');`	`select(Student).where(Student.city.in_(["Hyderabad", "Bangalore"]))`
`SELECT * FROM students WHERE phone IS NULL;`	`select(Student).where(Student.phone.is_(None))`
`SELECT * FROM students WHERE phone IS NOT NULL;`	`select(Student).where(Student.phone.is_not(None))`
`SELECT * FROM students JOIN departments ON students.department_id = departments.id;`	`select(Student).join(Department)`
`SELECT * FROM students LEFT JOIN departments ON students.department_id = departments.id;`	`select(Student).outerjoin(Department)`
`SELECT * FROM students ORDER BY age DESC LIMIT 5;`	`select(Student).order_by(Student.age.desc()).limit(5)`

Retrieving Results in SQLAlchemy

After building a query using select(), you need to decide how you want to retrieve the results. SQLAlchemy provides different methods depending on whether you expect one object, multiple objects, or a single value.

Result Retrieval Methods

Method	Returns	Use When
`all()`	List of objects	Retrieve all matching records.
`first()`	First object or `None`	Retrieve only the first matching record.
`one()`	Exactly one object	Expect exactly one record. Raises an exception if none or multiple records are found.
`one_or_none()`	One object or `None`	Expect at most one record. Raises an exception if multiple records are found.
`get()`	Object or `None`	Retrieve a record by its primary key.
`scalar()`	Single scalar value	Retrieve a single value such as `COUNT`, `SUM`, `AVG`, etc.
`scalars()`	ScalarResult	Retrieve ORM objects (or the first selected column) from a query. Usually followed by `all()`, `first()`, or `one()`.

`all()`

Returns all matching records.

students = session.scalars(select(Student)).all()

Result

[
    Student(...),
    Student(...),
    Student(...)
]

Use when you want every matching record.

`first()`

Returns only the first matching record.

student = session.scalars(select(Student)).first()

Result

Student(...)

None

Use when only the first matching record is required.

`one()`

Returns exactly one record.

student = session.scalars(
    select(Student).where(Student.id == 1)
).one()

Raises an exception if:

No record is found.
More than one record is found.

Use when you are certain exactly one record exists.

`one_or_none()`

Returns one record or None.

student = session.scalars(
    select(Student).where(Student.id == 1)
).one_or_none()

Returns

One object
None

Raises an exception if multiple records are found. Use when the record may or may not exist.

`get()`

Retrieves a record using its primary key.

student = session.get(Student, 1)

Returns

Student(...)

None

Use get() only when searching by the primary key.

`scalar()`

Returns a single value. Example:

count = session.scalar(
    select(func.count(Student.id))
)

Result

Commonly used with:

COUNT()
SUM()
AVG()
MIN()
MAX()

`scalars()`

Returns ORM objects (or the first selected column) from a query. Example

students = session.scalars(select(Student)).all()

Equivalent to

result = session.execute(select(Student))
students = result.scalars().all()

Use scalars() when retrieving ORM objects.

Quick Reference

Requirement	Method
All matching records	`.all()`
First matching record	`.first()`
Exactly one record	`.one()`
Zero or one record	`.one_or_none()`
Find by primary key	`session.get()`
Single value (COUNT, SUM, AVG…)	`session.scalar()`
ORM objects	`session.scalars()`

Choosing the Right Method

Need all matching records?
        │
       Yes
        │
        ▼
      all()

Need only the first record?
        │
       Yes
        │
        ▼
     first()

Searching by primary key?
        │
       Yes
        │
        ▼
   session.get()

Expect exactly one record?
        │
       Yes
        │
        ▼
      one()

Expect zero or one record?
        │
       Yes
        │
        ▼
   one_or_none()

Need a single value (COUNT, SUM...)?
        │
       Yes
        │
        ▼
     scalar()

Need ORM objects from a SELECT query?
        │
       Yes
        │
        ▼
    scalars()

ORM Relationships

In real-world applications, a single table is rarely enough to represent all the required data. Instead, multiple tables work together by establishing relationships between them. For example, in a Blog application:

A user can write multiple blog posts.
Every blog post belongs to one author.
A blog post can have multiple tags.
Every comment belongs to a blog post.

Instead of duplicating data across tables, relational databases connect tables using Primary Keys and Foreign Keys.

Why Relationships?

Suppose we store author information inside every blog post.

BlogPosts

+----+------------------+-----------+----------------+
| id | title            | author    | email          |
+----+------------------+-----------+----------------+
| 1  | FastAPI Basics   | John      | [email protected]  |
| 2  | SQLAlchemy ORM   | John      | [email protected]  |
+----+------------------+-----------+----------------+

This causes several problems.

Duplicate data
Wasted storage
Difficult updates
Risk of inconsistent data

Instead, store author information only once.

Users

+----+---------+----------------+--------+
| id | username| email          | role   |
+----+---------+----------------+--------+
| 1  | john    | [email protected]  | AUTHOR |
+----+---------+----------------+--------+

BlogPosts

+----+------------------+-----------+
| id | title            | author_id |
+----+------------------+-----------+
| 1  | FastAPI Basics   |     1     |
| 2  | SQLAlchemy ORM   |     1     |
+----+------------------+-----------+

Now the author details exist only once, making the database normalized and easier to maintain.

Primary Key vs Foreign Key

Every table has a Primary Key, which uniquely identifies each row. A Foreign Key stores the primary key value of another table.

Users
+----+----------+
| id | username |
+----+----------+
| 1  | john     |
| 2  | alice    |
+----+----------+

        ▲
        │
        │ Foreign Key
        │

BlogPosts
+----+----------------+-----------+
| id | title          | author_id |
+----+----------------+-----------+
| 1  | FastAPI Intro  |     1     |
| 2  | SQLAlchemy ORM |     1     |
| 3  | JWT Auth       |     2     |
+----+----------------+-----------+

Here,

Users.id is the Primary Key.
BlogPosts.author_id is the Foreign Key.

Types of Relationships

One-to-One (1:1)

One record is associated with exactly one record in another table. Examples

User → Profile
Employee → Passport
Student → Identity Card

User -------- Profile

One-to-Many (1:N)

One record can have multiple related records. This is the most common relationship. Example One author can write many blog posts.

John

├── FastAPI Basics
├── SQLAlchemy ORM
├── JWT Authentication
└── Docker Deployment

Only the child table stores the foreign key.

Many-to-One (N:1)

This is simply the reverse direction of One-to-Many. Many blog posts belong to one author.

Blog Post
      │
      ▼
    Author

Although conceptually different, SQLAlchemy implements One-to-Many and Many-to-One together.

Many-to-Many (M:N)

Many records on one side are related to many records on the other side. Example

Blog Post
      ▲
      │
 Association Table
      │
      ▼
      Tag

Examples

Students ↔ Courses
Users ↔ Roles
Blog Posts ↔ Tags
Movies ↔ Actors

Many-to-Many relationships always require an association table.

Relationship in Our Blog Application

Our application contains two tables.

User

id
username
email
role

        │
        │ One User
        ▼

BlogPost

id
title
content
author_id

One author can create multiple blog posts.

John

├── Blog 1
├── Blog 2
├── Blog 3
└── Blog 4

The relationship is therefore

One User → Many Blog Posts

User Roles

Our application stores both Authors and Admins inside the same users table.

Users

+----+----------+--------+
| id | username | role   |
+----+----------+--------+
| 1  | john     | AUTHOR |
| 2  | admin    | ADMIN  |
+----+----------+--------+

The role determines permissions. Author

Create posts
Update own posts
Delete own posts

Admin

View all posts
Update any post
Delete any post
Manage users

Notice that roles do not affect the database relationship. Both Admin and Author are simply users. The relationship remains

User 1 ------< BlogPosts

Authorization is handled by the application logic, not by the database relationship.

Relationship in SQLAlchemy

SQLAlchemy models relationships in two different ways.

Database Relationship

The database only understands foreign keys.

author_id = mapped_column(
    ForeignKey("users.id")
)

This creates the actual database constraint.

ORM Relationship

Python objects need an object reference. For that, SQLAlchemy provides relationship().

posts = relationship(
    "BlogPost",
    back_populates="author"
)

and

author = relationship(
    "User",
    back_populates="posts"
)

Now SQLAlchemy understands that these two models are connected.

Complete User Model

class User(Base):
    __tablename__ = "users"

    id = mapped_column(primary_key=True)
    username = mapped_column(unique=True)
    email = mapped_column(unique=True)
    role = mapped_column()

    posts = relationship(
        "BlogPost",
        back_populates="author"
    )

Complete BlogPost Model

class BlogPost(Base):
    __tablename__ = "blog_posts"

    id = mapped_column(primary_key=True)
    title = mapped_column()
    content = mapped_column()

    author_id = mapped_column(
        ForeignKey("users.id")
    )

    author = relationship(
        "User",
        back_populates="posts"
    )

How SQLAlchemy Uses Relationships

Unlike SQL, SQLAlchemy lets us navigate related objects directly. Instead of writing joins, we simply access attributes.

Get all posts written by a user

user.posts

Returns

[
    BlogPost(...),
    BlogPost(...),
    BlogPost(...)
]

Get the author of a blog post

post.author

Returns

User(...)

Creating Relationships

Instead of assigning the foreign key manually,

post.author_id = user.id

you can assign the object itself.

post.author = user

SQLAlchemy automatically updates the foreign key during session.commit(). Similarly,

user.posts.append(post)

automatically sets

post.author_id

This synchronization is handled by the ORM.

Understanding `back_populates`

back_populates connects both relationship properties.

User.posts
      ▲
      │
back_populates
      │
      ▼
BlogPost.author

Without back_populates, each relationship behaves independently. With back_populates:

Updating one side updates the other.
Both objects remain synchronized.
Navigation works in both directions.

Database Relationship vs ORM Relationship

Database	SQLAlchemy ORM
Foreign Key	relationship()
Enforced by Database	Managed by SQLAlchemy
Stores IDs	Stores Object References
Used by SQL	Used by Python

Both work together. The database guarantees referential integrity, while SQLAlchemy provides convenient object navigation.

Practice

To reinforce what you’ve learned in this section, practice with the interactive follow-along notebook:

Follow-Along Practice

Practice creating SQLAlchemy engines, declaring ORM Base models, running CRUD operations, and building relationships between tables.💻 VS Code | 🚀 Colab | 📥 Download

Summary

Relationships connect tables using foreign keys.
Primary keys uniquely identify rows.
Foreign keys reference another table’s primary key.
SQLAlchemy models relationships using relationship().
One-to-Many is the most common relationship in REST APIs.
back_populates connects both sides of the relationship.
SQLAlchemy lets you navigate relationships using Python objects instead of writing SQL joins.
User roles (Author/Admin) determine permissions but do not change the database relationship.
The database manages data integrity, while SQLAlchemy provides an object-oriented interface to work with related data.

​Learning Objectives

​Topics Covered

​Working Without an ORM

​Working With an ORM

​What is an ORM?

​The Big Picture

​Writing Data (INSERT / UPDATE / DELETE)

​Reading Data (SELECT)

​Understanding the Components

​Engine

​Session

​ORM Model

​SQLAlchemy

​Database

​Putting Everything Together

​Summary

​Building Your First SQLAlchemy ORM Application

​Learning Objectives

​Project Setup

​Step 1 — Create the Engine

​Workflow

​Step 2 — Create the Declarative Base

​Workflow

​Step 3 — Create an ORM Model

​Mapping

​Step 4 — Create Database Tables

​What happens?

​Step 5 — Create a Session

​Workflow

​Step 6 — Insert a Record

​Workflow

​Complete Application

​Summary

​CRUD Operations with SQLAlchemy ORM

​Create (INSERT)

​SQL

​SQLAlchemy ORM

​Workflow

​Read (SELECT)

​Retrieve All Records

​SQL

​SQLAlchemy ORM

​Retrieve by Primary Key

​SQL

​SQLAlchemy ORM

​Retrieve the First Record

​SQL

​SQLAlchemy ORM

​Workflow

​Update (UPDATE)

​Step 1 – Retrieve the Object

​Step 2 – Modify the Object

​Step 3 – Save the Changes

​Complete Example

​SQL

​SQLAlchemy ORM

​Workflow

​Delete (DELETE)

​Step 1 – Retrieve the Object

​Step 2 – Delete the Object

​Step 3 – Save the Changes

​Complete Example

​SQL

​SQLAlchemy ORM

​Workflow

​CRUD Summary

​Complete CRUD Example

​SQL SELECT vs SQLAlchemy ORM

​Retrieving Results in SQLAlchemy

​Result Retrieval Methods

​all()

​first()

​one()

​one_or_none()

​get()

​scalar()

​scalars()

​Quick Reference

​Choosing the Right Method

​ORM Relationships

Learning Objectives

Topics Covered

Working Without an ORM

Working With an ORM

What is an ORM?

The Big Picture

Writing Data (INSERT / UPDATE / DELETE)

Reading Data (SELECT)

Understanding the Components

Engine

Session

ORM Model

SQLAlchemy

Database

Putting Everything Together

Summary

Building Your First SQLAlchemy ORM Application

Learning Objectives

Project Setup

Step 1 — Create the Engine

Workflow

Step 2 — Create the Declarative Base

Workflow

Step 3 — Create an ORM Model

Mapping

Step 4 — Create Database Tables

What happens?

Step 5 — Create a Session

Workflow

Step 6 — Insert a Record

Workflow

Complete Application

Summary

CRUD Operations with SQLAlchemy ORM

Create (INSERT)

SQL

SQLAlchemy ORM

Workflow

Read (SELECT)

Retrieve All Records

SQL

SQLAlchemy ORM

Retrieve by Primary Key

SQL

SQLAlchemy ORM

Retrieve the First Record

SQL

SQLAlchemy ORM

Workflow

Update (UPDATE)

Step 1 – Retrieve the Object

Step 2 – Modify the Object

Step 3 – Save the Changes

Complete Example

SQL

SQLAlchemy ORM

Workflow

Delete (DELETE)

Step 1 – Retrieve the Object

Step 2 – Delete the Object

Step 3 – Save the Changes

Complete Example

SQL

SQLAlchemy ORM

Workflow

CRUD Summary

Complete CRUD Example

SQL SELECT vs SQLAlchemy ORM

Retrieving Results in SQLAlchemy

Result Retrieval Methods

`all()`

`first()`

`one()`

`one_or_none()`

`get()`

`scalar()`

`scalars()`

Quick Reference

Choosing the Right Method

ORM Relationships