🧠 ipykernel + Conda + PyCharm — Complete Guide

A comprehensive reference for managing Jupyter kernels across conda environments and using them in PyCharm. Perfect for ML/DL practitioners who work with multiple conda environments.

---

📋 Table of Contents

• Prerequisites
• Installing ipykernel
• Creating Custom Kernels
• Listing Kernels
• Removing Kernels
• Restoring Kernels
• Using Kernels in PyCharm
• Using Kernels in Jupyter Notebook/Lab
• Managing Multiple Environments
• Troubleshooting
• Quick Reference Cheatsheet

---

Prerequisites

• Anaconda or Miniconda installed
• At least one conda environment created
• PyCharm (Professional or Community) installed

Create a conda environment (if not done yet)

conda create -n my_env python=3.10
conda activate my_env

---

Installing ipykernel

ipykernel must be installed inside each conda environment you want to use as a kernel.

Install in current active environment

conda activate my_env
conda install ipykernel

Install via pip (alternative)

conda activate my_env
pip install ipykernel

Install in a specific environment without activating

conda install -n my_env ipykernel

---

Creating Custom Kernels

After installing ipykernel, register the environment as a Jupyter kernel.

Basic kernel registration

conda activate my_env
python -m ipykernel install --user --name my_env --display-name "My Environment"

Parameters explained

Parameter	Description
--user	Installs kernel for current user only (recommended)
--name	Internal kernel name (no spaces, used in commands)
--display-name	Human-readable name shown in PyCharm/Jupyter UI

Real-world examples

# ML environment
conda activate ML_Unsupervised
python -m ipykernel install --user --name ML_Unsupervised --display-name "Conda (ML Unsupervised)"

# Deep Learning environment
conda activate DL_Architecture_Generative
python -m ipykernel install --user --name DL_Generative --display-name "Conda (DL Generative)"

# Data Engineering environment
conda activate data_eng
python -m ipykernel install --user --name data_eng --display-name "Conda (Data Engineering)"

✅ Registered kernels are permanent — they survive terminal closes, system restarts, and PyCharm restarts.

---

Listing Kernels

List all registered kernels

jupyter kernelspec list

Example output

Available kernels:
  ml_unsupervised    C:\Users\username\AppData\Roaming\jupyter\kernels\ml_unsupervised
  dl_generative      C:\Users\username\AppData\Roaming\jupyter\kernels\dl_generative
  python3            C:\Users\username\anaconda3\share\jupyter\kernels\python3

Kernel storage locations

OS	Default kernel path
Windows	C:\Users\username\AppData\Roaming\jupyter\kernels\
macOS	~/Library/Jupyter/kernels/
Linux	~/.local/share/jupyter/kernels/

---

Removing Kernels

Remove a specific kernel

jupyter kernelspec remove kernel_name

Remove multiple kernels at once

jupyter kernelspec remove kernel1 kernel2 kernel3

Real-world examples

# Remove default python3 kernel
jupyter kernelspec remove python3

# Remove a custom kernel
jupyter kernelspec remove ml_unsupervised

# Remove multiple at once
jupyter kernelspec remove python3 dl_generative

⚠️ You will be prompted [y/N] — type y to confirm each removal.

💡 Removing a kernel does not delete the conda environment — only its Jupyter registration.

---

Restoring Kernels

Removed kernels can always be re-registered with the same command used to create them.

Restore python3 kernel

conda activate my_env
python -m ipykernel install --user --name python3 --display-name "Python 3"

Restore any custom kernel

conda activate ML_Unsupervised
python -m ipykernel install --user --name ML_Unsupervised --display-name "Conda (ML Unsupervised)"

---

Using Kernels in PyCharm

Step 1 — Set the Python Interpreter

1. Go to File → Settings (Ctrl+Alt+S)
2. Navigate to Project → Python Interpreter
3. Click the dropdown → Add Interpreter → Add Local Interpreter
4. Choose System Interpreter
5. Browse to your conda env's python.exe:

   # Windows
   C:\Users\username\anaconda3\envs\my_env\python.exe
   
   # macOS/Linux
   /Users/username/anaconda3/envs/my_env/bin/python
   

6. Click OK

Step 2 — Configure Jupyter Server

1. Go to File → Settings → Languages & Frameworks → Jupyter → Jupyter Servers
2. Select "IDE-Managed Server" (set to Auto)
3. Click OK

✅ PyCharm will auto-start/stop Jupyter — no manual commands needed.

Step 3 — Select Kernel in Notebook

1. Open any .ipynb file in PyCharm
2. Look at the top-right corner of the notebook editor
3. Click the kernel dropdown
4. Select your registered kernel (e.g. Conda (ML Unsupervised))

Switching kernels mid-session

• Top-right dropdown → select a different kernel
• PyCharm will restart the kernel automatically

---

Using Kernels in Jupyter Notebook/Lab

Start Jupyter from any environment

conda activate base
jupyter notebook
# or
jupyter lab

Change kernel in a notebook

• Jupyter Notebook: Kernel menu → Change Kernel → select your env
• JupyterLab: Click kernel name in top-right → select your env

Serve all conda envs automatically (optional)

Install nb_conda_kernels in base to auto-detect all conda environments:

conda install -n base nb_conda_kernels

Then start Jupyter from base — all envs appear as kernels automatically.

---

Managing Multiple Environments

Recommended workflow for multiple projects

# Create environments
conda create -n project_ml python=3.10
conda create -n project_dl python=3.11

# Install ipykernel in each
conda activate project_ml
conda install ipykernel
python -m ipykernel install --user --name project_ml --display-name "Project ML"

conda activate project_dl
conda install ipykernel
python -m ipykernel install --user --name project_dl --display-name "Project DL"

# Verify
jupyter kernelspec list

Check which kernel a notebook is using

The kernel name is stored in the notebook's metadata — visible in the top-right of PyCharm/Jupyter.

---

Troubleshooting

Kernel not showing in PyCharm

# Re-register the kernel
conda activate my_env
python -m ipykernel install --user --name my_env --display-name "My Env"
# Then restart PyCharm

Jupyter connection refused (port 8888)

# Check if Jupyter is running
jupyter notebook list

# Kill all running Jupyter servers
jupyter notebook stop

Wrong Python version in kernel

# Verify which Python the kernel uses
conda activate my_env
python --version
# Then re-register
python -m ipykernel install --user --name my_env --display-name "My Env (Python 3.10)"

Kernel keeps dying/crashing

# Check if all required packages are installed in the env
conda activate my_env
conda list

# Reinstall ipykernel
conda install --force-reinstall ipykernel

PyCharm shows "No running Jupyter servers found"

• Go to Settings → Jupyter → Jupyter Servers
• Switch to "IDE-Managed Server"
• This makes PyCharm handle the server automatically

Duplicate kernels appearing

# List all kernels to find duplicates
jupyter kernelspec list

# Remove duplicates
jupyter kernelspec remove duplicate_kernel_name

---

Quick Reference Cheatsheet

# Installing ipykernel
conda activate my_env
conda install ipykernel

# creating kernel servers
python -m ipykernel install --user --name env_name --display-name "UI_display_name"

# listing all kernels
jupyter kernelspec list

# removing kernels
jupyter kernelspec remove kernel_name            # remove one
jupyter kernelspec remove k1 k2 k3              # remove multiple

# restoring kernels
python -m ipykernel install --user --name env_name --display-name "UI_display_name"

# Jupyter Notebook commands
jupyter notebook list                            # list running servers
jupyter notebook stop                            # stop all servers
jupyter lab                                      # start JupyterLab

---

🤝 Contributing

Feel free to open issues or pull requests for additional tips, OS-specific instructions, or new troubleshooting entries.

---

📄 License

MIT License — free to use, share, and modify.

---