I implement most algorithms I develop into Julia packages.
For generic atomistic simulation I used to maintain JuLIP.jl but this is no longer maintained and instead I try to contribute to JuliaMolSim; see also the landing page. There is now a small Julia Chemistry community developing.
Most of my research these days is centered around developing machine learning surrogates for particle systems using the atomic cluster expansion framework. Most codes that I develop are maintained within the ACEsuit github org, including ML methods for interatomic potentials, hamiltonians, etc, and more.
ACEpotentials.jl: a reasonably mature package for fitting linear MLIPs (in theory nonlinear as well but this functionality is not well developed yet)
ACEhamiltonians.jl: fitting linear ML tight binding models
EquivariantTensors.jl: an attempt to implement general performant Lux.jl "layers" from which linear and nonlinear ACE and related models can be built, to become the backend for most ACEsuit packages.
But there are several more experimental packages, for CGMD, wave functions and more.
SaddleSearch.jl : experimental library for saddle point search algorithms, including walker and string methods
SlaterKoster.jl : implements the Slater-Koster transformation as the basis of tight-binding models
NBodyIPs.jl : interatomic potentials for materials based on the PIP formalism
JuLIPMaterials.jl : code collection for atomistic material simulation, e.g., elasticity, boundary conditions, ...
Isaac.jl : Experimental Newton-Krylov methods for minimisation and saddle point search
TightBinding.jl : an experimental tight-binding code with some unique features, in particular it supports site energies for QM/MM multi-scale methods
See my github page for more.
(many years ago...) I collected some Julia example/tutorial notebooks that I wrote to teach myself Julia. They are intended primarily as a rapid introduction to Julia for numerical analysts, mostly from the point of view of differential equations, and most useful for somebody familiar with Matlab. For more details see the github repository.