diff --git a/.gitignore b/.gitignore
index c8b92581..a5b84e97 100644
--- a/.gitignore
+++ b/.gitignore
@@ -51,6 +51,9 @@ venv.bak/
 # Jupytext: we version converted md files not ipynb sources
 *.ipynb
 
+# Jupyter
+.ipynb_checkpoints/
+
 # -- Provisioning
 provisioning/.terraform*
 provisioning/terraform.tfstate*
diff --git a/docker-compose.yml b/docker-compose.yml
index fc1b53d0..ee7f7895 100644
--- a/docker-compose.yml
+++ b/docker-compose.yml
@@ -80,6 +80,8 @@ services:
       NB_GID: ${DOCKER_GID:-1000}
       CHOWN_HOME: 'yes'
       CHOWN_HOME_OPTS: -R
+    env_file:
+      - env.d/notebook
     ports:
       - 8888:8888
     volumes:
diff --git a/env.d/notebook b/env.d/notebook
new file mode 100644
index 00000000..f67cbce7
--- /dev/null
+++ b/env.d/notebook
@@ -0,0 +1 @@
+DATABASE_URL=postgresql+psycopg://qualicharge:pass@postgresql:5432/qualicharge-api
diff --git a/src/notebook/Dockerfile b/src/notebook/Dockerfile
index e6f1091d..b00926c6 100644
--- a/src/notebook/Dockerfile
+++ b/src/notebook/Dockerfile
@@ -1,11 +1,15 @@
 # -- Custom image --
-FROM jupyter/base-notebook
+FROM quay.io/jupyter/base-notebook:notebook-7.2.1
 
 # Install base dependencies
+#
+# FIXME: jupytext 1.16.4+ seems to fix the issue but is not released yet
+#        see: https://github.com/mwouts/jupytext/issues/1260
 RUN mamba install --yes \
       duckdb \
       geopandas \
-      jupytext \
+      jupytext==1.16.2 \
       matplotlib \
       pandas \
+      psycopg[binary,pool] \
       seaborn
diff --git a/src/notebook/example.md b/src/notebook/example.md
new file mode 100644
index 00000000..1310479b
--- /dev/null
+++ b/src/notebook/example.md
@@ -0,0 +1,145 @@
+---
+jupyter:
+  jupytext:
+    formats: ipynb,md
+    text_representation:
+      extension: .md
+      format_name: markdown
+      format_version: '1.3'
+      jupytext_version: 1.16.2
+  kernelspec:
+    display_name: Python 3 (ipykernel)
+    language: python
+    name: python3
+---
+
+# QualiCharge data: an example notebook
+
+This notebook aims to be an example notebook used as a starting point for a new analysis or indicator calculation. It provides code snippets and examples to fetch and record data from our PostgreSQL database.
+
+## Create the database engine
+
+```python
+import os
+from sqlalchemy import create_engine
+
+# Get database URL from the environment
+database_url = os.getenv("DATABASE_URL")
+
+# Create a database engine that will be used to generate connections
+engine = create_engine(database_url)
+```
+
+## Fetch data from the database
+
+### Example 1: generate a stations map using GeoPandas
+
+```python
+from geopandas import GeoDataFrame
+
+query = """
+SELECT
+  Station.nom_station,
+  Station.id_station_itinerance,
+  Amenageur.nom_amenageur as amenageur,
+  Localisation."coordonneesXY" as geom
+FROM
+  Station
+INNER JOIN Localisation ON Station.localisation_id = Localisation.id
+INNER JOIN Amenageur ON Station.amenageur_id = Amenageur.id
+"""
+
+
+with engine.connect() as conn:
+    # Query a PostgreSQL database using the PostGIS extension
+    stations = GeoDataFrame.from_postgis(query, conn)
+
+print(f"Loaded {len(stations.index)} stations")
+stations.sample(10)
+```
+
+```python
+# Display an interactive map of the stations
+stations.explore(column="amenageur")
+```
+
+### Example 2: explore operators distribution
+
+```python
+import pandas as pd
+
+query = """
+SELECT
+  Operateur.nom_operateur,
+  PointDeCharge.id_pdc_itinerance
+FROM
+  PointDeCharge
+  INNER JOIN Station ON PointDeCharge.station_id = Station.id
+  INNER JOIN Operateur ON Station.operateur_id = Operateur.id
+"""
+
+with engine.connect() as conn:
+    # Query a PostgreSQL database using the PostGIS extension
+    pdcs = pd.read_sql_query(query, conn)
+
+print(f"Loaded {len(pdcs.index)} points of charge")
+pdcs.sample(10)
+```
+
+```python
+import seaborn as sns
+
+# Render a barplot with the number of points of charge by operator
+sns.barplot(data=pdcs.value_counts("nom_operateur"))
+```
+
+## Write data to the database
+
+### Example 1: create a new table with calculated indicator
+
+In this example, we will calculate the number of points of charge per French department at a particular date/time (now) and store this stateful snapshot in the database.
+
+```python
+import uuid
+import pandas as pd
+
+# Get the city code insee for each point of charge
+query = """
+SELECT
+  Localisation.code_insee_commune
+FROM
+  PointDeCharge
+  INNER JOIN Station ON PointDeCharge.station_id = Station.id
+  INNER JOIN Localisation ON Station.localisation_id = Localisation.id
+"""
+with engine.connect() as conn:
+    # Query a PostgreSQL database using the PostGIS extension
+    codes_insee = pd.read_sql_query(query, conn)
+
+# Add a department column
+codes_insee["department"] = codes_insee["code_insee_commune"].map(lambda x: int(x[:2]) if x else None)
+
+# Calculate our indicator and add a timestamp to each department counts (row)
+indicator = codes_insee.value_counts("department").to_frame().reset_index()
+indicator["calculated_at"] = pd.Timestamp.now()
+
+# Set UUIDs as the index
+indicator["uuid"] = indicator.apply(lambda _: uuid.uuid4(), axis=1)
+indicator.set_index("uuid", inplace=True)
+
+# Explictly set the department column as integers
+indicator = indicator.astype({"department": "int32"})
+indicator
+```
+
+```python
+# Save the indicator to a (new) table
+indicator.to_sql("IDepartmentDynamic", engine, if_exists="append")
+```
+
+```python
+# Check inserted results
+query = 'SELECT * FROM "IDepartmentDynamic" WHERE department = 75'
+paris = pd.read_sql_query(query, engine)
+paris
+```