creating Venv

 

Data Engineer Virtual Environment (venv) + Cloud Workflow Notes

1. What is a Virtual Environment (venv)?

A virtual environment is an isolated Python environment for a project.

It allows each project to have its own Python libraries and versions.

Example:

Project A
|
└── venv
    ├── pandas 2.0
    └── pyspark


Project B
|
└── venv
    ├── pandas 2.2
    └── tensorflow

Without venv, package versions can conflict.

---

2. Why Data Engineers need venv

Data engineers use many Python libraries:

• pandas

• numpy

• pyspark

• requests

• cloud SDKs

Example:

Automation script:

Python script
      |
      |
      v
Azure Storage / GCP Storage / APIs

The venv keeps the required libraries separate.

---

3. venv in Cloud Environment

Cloud does NOT replace venv.

Cloud = infrastructure

venv = Python dependency isolation

Example:

Cloud VM
|
├── Python
|
├── Project A
|     └── venv
|
└── Project B
      └── venv

---

4. Team Environment: Do we share the same venv?

NO.

Each developer has their own venv.

Example:

Developer A laptop
|
└── venv


Developer B laptop
|
└── venv

The team shares:

Git Repository

├── Python code
├── requirements.txt
└── configuration

The team does NOT share:

❌ venv folder

---

5. requirements.txt

This file stores package dependencies.

Example:

pandas==2.2.0
numpy==1.26.0
pyspark==3.5.0
requests==2.32.0

Team member creates the same environment:

pip install -r requirements.txt

---

6. Docker vs venv

venv

Used mainly for:

• local development

• testing Python scripts

Example:

Laptop

venv
 |
Python libraries
 |
Code

---

Docker

Used for:

• production deployment

• cloud environments

Example:

Docker Container

Python
 |
Libraries
 |
Application

Usually:

Developer:
venv

Production:
Docker

---

7. Azure Data Engineer Example

Stack:

• Azure Data Factory

• Azure Databricks

• Power BI

Flow:

Developer Laptop

venv
 |
Python automation scripts

        |
        v

Azure DevOps Git

        |
        v

ADF / Databricks / Power BI

---

venv in Azure Data Engineering

Used for:

✅ Python automation scripts
✅ Azure SDK scripts
✅ Data validation scripts
✅ Testing ETL logic

Not used for:

❌ ADF itself
❌ Power BI
❌ Databricks cluster runtime

---

8. GCP Data Engineer Example

Stack:

• Cloud Composer

• Dataflow

• Dataproc

• BigQuery

Flow:

Developer

venv

Python code

Git

CI/CD

Cloud Services

venv used for:

✅ Google Cloud SDK scripts
✅ Apache Beam testing
✅ BigQuery automation
✅ Local development

Not used for:

❌ BigQuery engine
❌ Dataproc cluster Python environment

---

9. Creating a venv (Windows)

Create:

python -m venv firstenv

Meaning:

python
 |
 -m = run module
 |
 venv = Python virtual environment module
 |
 firstenv = environment name

---

10. Activate venv (Windows CMD)

Correct:

firstenv\Scripts\activate

You should see:

(firstenv) C:\Users\keert>

---

Deactivate:

deactivate

---

11. Activate venv (Linux/Mac)

source firstenv/bin/activate

---

12. Important venv Commands

Create environment

python -m venv venv

---

Activate

Windows:

venv\Scripts\activate

Linux/Mac:

source venv/bin/activate

---

Exit

deactivate

---

Check Python version

python --version

---

Check which Python is running

Windows:

where python

Should show:

venv\Scripts\python.exe

---

13. pip Commands

Install package

pip install pandas

---

Install project dependencies

pip install -r requirements.txt

Meaning:

-r = read requirements file

---

List libraries

pip list

---

Show exact versions

pip freeze

Example:

pandas==2.2.0
numpy==1.26.0

---

Save environment

pip freeze > requirements.txt

Team shares this file.

---

Remove package

pip uninstall pandas

---

Upgrade package

pip install --upgrade pandas

or:

pip install -U pandas

---

Package details

pip show pandas

---

14. Meaning of Common Flags

Flags start with -.

Python

-m

Run module:

python -m venv venv

---

pip

-r

Read requirements file:

pip install -r requirements.txt

---

-U

Upgrade:

pip install -U pandas

---

--help

Help:

pip --help

---

15. Other Package Managers

Python packages

pip

Example:

pip install pandas

---

Windows software

Chocolatey:

choco install git

Winget:

winget install python

---

Mac software

Homebrew:

brew install python
brew install git

---

16. Data Engineer Daily Workflow

Typical:

git clone project

cd project

python -m venv venv

venv\Scripts\activate

pip install -r requirements.txt

python script.py

pip freeze > requirements.txt

deactivate

---

Most Important Commands to Memorize

python -m venv venv

venv\Scripts\activate

deactivate

where python

pip install package

pip list

pip freeze

pip freeze > requirements.txt

pip install -r requirements.txt

pip uninstall package

---

Final Mental Model

venv
 =
Developer's Python workspace


requirements.txt
 =
Team dependency sharing


Docker
 =
Production environment


Git
 =
Code sharing


Azure DevOps/GCP CI-CD
 =
Automation


ADF / Composer
 =
Orchestration


Databricks / Dataflow
 =
Processing


Power BI / BigQuery
 =
Analytics layer

This is the workflow you should remember as a Data Engineer.

step 1. Make sure the python is installed in system.

step 2 . make sure the python is updated to python version 3

or give command -m pip install --upgrade pip(pip is the python package manager aND CHOCO is the windows package manager manager and brew is for MAC )

step 3 : create the venv using command

python -m venv firstenv

python → runs Python

-m → tells Python "run a module"

venv → the module

firstenv → name of your environment

step 4 : Go inside Venv.

go inside the path and give the command 

firstenv\Scripts\activate

and its shows like 

(firstenv) C:\Users\keert>pip install pandas ( install the libraies which we need to develops)

step 5 : To know what packages or libraies installed give the command

(firstenv) C:\Users\keert> pip freeze

(firstenv) C:\Users\keert> pip list 

step 6 : In gendral the teams give the list of requiremnts in git for examples in requirements.txt 

for that we need to use pip install -r requirements.txt

step 7 : to come out of the environment give command 

(firstenv) C:\Users\keert>deactivate.

to inside again give same C:\Users\keert>firstenv\scripts\activate

do the work we want.

• The Cloud vs. venv Distinction: This is a massive stumbling block for beginners. Recognizing that Cloud = Infrastructure and venv = Python Isolation shows you understand that running an EC2, VM, or Cloud Function doesn't magically solve dependency hell.

• Databricks/Dataproc Runtimes: You correctly noted that venv is not used for managed spark cluster runtimes. Databricks handles its own cluster-level libraries (or uses init scripts/%pip), so keeping your local script automation isolated from cluster runtimes is exactly how it works in production.

• The Git Guardrail: Omitting the ❌ venv folder from Git and only sharing requirements.txt is standard industry best practice. (Pro-tip: Always add venv/ or .venv/ to your .gitignore file immediately).

• The "where python" / "which python" Check: Checking your environment context via where python (Windows) or which python (Mac/Linux) is the ultimate debugging step when a package isn't loading properly.

🔍 Micro-Tips to Level Up Your Notes

Your commands and logic are flawless, but as you work in enterprise environments, keep these tiny nuances in mind:

1. Deterministic Builds (pip-compile): While pip freeze > requirements.txt is perfect, it dumps everything, including sub-dependencies. As you advance, you might encounter tools like pip-tools or Poetry which separate your top-level dependencies (e.g., just pandas) from their underlying sub-dependencies to keep things cleaner.

2. Mac/Linux Equivalent for where: You noted where python for Windows. For Mac/Linux, the exact equivalent command to memorize is:

   Bash

   which python
   

3. Docker + venv (The Hybrid Approach): While your distinction that Developer = venv and Production = Docker is great for a mental model, in advanced production environments, engineers actually use both together. Inside a Dockerfile, it is common practice to create a venv to isolate the app from the base OS Python layer.