Functions - get comfortable writing your own functions for repeatable tasks.
Github - any coder should use some kind of source control, R’s integration with Github makes it a natural fit.
Create a package - it doesn’t have to appear in CRAN (yet!) but its a small step from sourcing a file of your custom functions to loading your own private library.
Shiny - being able to quickly create interactive webapps when asked for a report is always impressive
Kaggle - practice your machine learning with public datasets in a friendly competitive environment
Advanced packages - get a package on CRAN, or contribute to the development of one on Github. Start using tests, Travis and coverage
Start using unit tests via library(testthat) to ensure robust code.