An image built off of jupyter/base-notebook that includes the free Wolfram Engine for developers.

docker build -t jupyter-mathematica .

docker run -p 8888:8888 jupyter-mathematica

Then inside the container, run wolframscript and login with your Wolfram username and password or get a free license here. You can easily do this from within a shell in Jupyter. After you are logged in, run

$PasswordFile // FilePrint

in the wolframscript session to get your license key. Create an empty directory somewhere on your host system named license. Copy and paste the output from the wolframscript command into a file named mathpass within that directory. We will mount this directory into the container when starting it from now on. Example:

docker run -p 8888:8888 \
-v /home/$USER/jm-files/license:/usr/local/Wolfram/WolframEngine/12.1/Configuration/Licensing \
jupyter-mathematica

Now, we need to tell Jupyter about mathematica. Run the following commands in the container.

cd /home/jovyan/wolfram/WolframLanguageForJupyter
./configure-jupyter.wls add
jupyter-kernelspec list

The command jupyter-kernelspec list should tell you which kernels are available, make sure you see the kernel we just added in the output. Now, you should be able to start a Wolfram Language notebook. Keep in mind that any notebooks you create will be stored in temporary docker volumes. Consider bind mounting some local directory like this:

docker run -p 8888:8888 -v /home/$USER/jm-files/work:/home/jovyan/work jupyter-mathematica

Full example mounting license file and work directory:

docker run -p 8888:8888 -v /home/$USER/jm-files/work:/home/jovyan/work \
-v /home/$USER/jm-files/license:/usr/local/Wolfram/WolframEngine/12.1/Configuration/Licensing \
jupyter-mathematica

If the WolframLanguage Kernel doesn't appear, you may have to re-run configure-jupyter.wls add and then jupyter-kernelspec list. I have experienced this behavior periodically and when restarting the container. After running this command, wait a few secodns and the kernel should appear.