SmartHire is an open source applicant prioritization system powered by machine learning. The system was created to help organizations go through large backlogs of applicants and find the best candidates for their open positions. The system learns who the best candidates are based on who has previously made it through the hiring process. SmartHire is built on Node.js, AngularJS, Elasticsearch, Apache Spark, Apache Tika, Apache OpenNLP and Scala.
SmartHire is powered by a number of dependencies. Npm and sbt help with managing most of these, but some preliminary setup is required. Instructions are provided below:
SmartHire uses Node as its web server. The download page for Node can be found here. For instructions on installing node via a package manager, see here.
Npm (node package manager) is used to install the dependencies for the server. The newer releases of Node come with npm, so be sure to check that it is included.
Elasticsearch is used to store and quickly serve resume data. Install instructions can be found on the Elasticsearch website here.
Sbt, or scala build tool, is required for downloading the dependencies of the ETL processes. Its download page lives here.
Spark is a library that allows SmartHire to scale to any number of resumes. It is located here. It is recommended to pick up spark version 1.6.1.
If any of these programs are installed manually, make sure to set up the proper environment variables in a shell startup script. For a user who installs in the home folder, configuration may look something like this...
export SPARK_HOME=/home/*USER-NAME*/spark-1.6.1-bin-hadoop2.6
export PATH=$PATH:/home/*USER-NAME*/elasticsearch-2.3.3/bin
export PATH=$PATH:$SPARK_HOME/binNow that the arduous installation task is over, it is time to begin firing everything up. This will be accomplished in two main steps. The first is to set up all of the required servers. Details on how this can be done are located in the app directory README. The second step is to load data for the app to use. This process is described in the etl directory README.
SmartHire was built by the Data Works 2016 summer internship team in response to ... the hiring of the Data Works 2016 summer internship team. While designed primarily for the task of filtering future Data Works interns, it is hoped that the application is extensible enough to be useful to others as well.
Alissa Cobb
Brantley Gilbert
Dennis Huang
Nitin Sudini
William Goubeaux