**Update 20190705**

Given the popularity of this question, it might be good to know why the OP (and possibly many others) didn't get WL running under Jupyter. So I provide the recipe below (all below tested under macOS):

### 1. Inspect the installed kernels:

By running 
```
jupyter kernelspec list
```

One would see:

```
python3                /Users/sunt05/Library/Jupyter/kernels/python3
wlengine               /Users/sunt05/Library/Jupyter/kernels/wlengine
wolframlanguage12      /Users/sunt05/Library/Jupyter/kernels/wolframlanguage12
```
So we know the kernels are correctly installed (here I have both Mathematica and WLE).

### 2. Explore the kernel spec file:

By navigating into one of the above folders, one may see the magic happens right in the `kernel.json` file.

The one below is for WL:
```json
{
	"argv":[
		"/Applications/Wolfram Engine.app/Contents/Resources/Wolfram Player.app/Contents/MacOS/WolframKernel",
		"-script",
		"\/Users\/sunt05\/Downloads\/WolframLanguageForJupyter-master\/WolframLanguageForJupyter\/Resources\/KernelForWolframLanguageForJupyter.wl",
		"{connection_file}",
		"ScriptInstall"
	],
	"display_name":"Wolfram Language 12 WL",
	"language":"Wolfram Language"
}

```

So if the kernel spec is not properly installed, one can also manually create a such kernel json file and put the info of WLE in.

Hope this might be helpful.

---

Simply install the kernel for the Jupyter environment following the official guide: https://github.com/WolframResearch/WolframLanguageForJupyter


then WL would be there (tested on macOS with Anaconda Python 3.7.3):


[![enter image description here][1]][1]


A noting caveat is that the auto-completion is missing.

Environment info:
```
conda version : 4.6.14
conda-build version : 3.17.8
python version : 3.7.3.final.0
```




  [1]: https://i.sstatic.net/BK5uU.png