I had an idea in mind, but in order to prove it would work I needed some sample data.  After much digging I found various places on the internet where I could get resumes.  Lots of them.  After some hard work and trial and error the first prototype was born.  By prototype, I mean a database that I could run queries on.  Even still, I knew exactly what it represented and what potential it had.  With data that represents individual transitions from one job to another, I had built an interconnected web of jobs — more technically speaking, a weighted graph:

Jobs = Nodes
Transitions = Edges

This graph could be searched for optimal paths between jobs.  It could be used to predict what was the likely job following any given job and what was the likely path prior to that job.  Then I created a second weighted graph to map the relationship between college degrees and jobs.  This powerful job graph is the core of JobTitled is today.  It allows us to programmatically perform career analytics on a large scale.