Update: A simpler alternative is to install nb_conda_kernels, which makes accessible all defined environments.
Two commands to run when creating a new env:
# 1. Env creation: always include ipykernel (could be place in .condarc)
conda create --name env310 python=3.10 ipykernel numpy
# 2. Kernel installation: makes env available in jupyter's kernels list
python -m ipykernel install --user --name=env310 --display-name "py310"