The problem
I built Sourcing Compass around an early search question: where might people have done work that matters for this role? I wanted a way to explore target companies, adjacent industries, and less obvious talent pools together.
Making that map visible gives a recruiter and hiring leader something to discuss, challenge, and refine before outreach.
What I built
Sourcing Compass is a one-page React app. It takes role context such as title, seniority, company context, key skills, and exclusions, then returns four structured categories:
- Target companies: the most likely sources, with fit scores and tags
- Adjacent talent pools: consulting firms, research labs, adjacent verticals that carry transferable experience
- Wildcard bets: non-obvious, high-signal sources for when the standard pools are exhausted
- Target titles: exact and variant job titles with match confidence scores, ready to paste into a search
Nodes are connected. Hover any card and dashed lines show which companies map to which titles, which adjacent pools feed which targets. The entire result set is a graph, not a list.
Why it matters
This gives a sourcing kick-off a first map of relevant organisations, adjacent talent pools, and related titles. I treat that map as a set of hypotheses to check with the hiring leader and against the market. It gives us something specific to challenge and refine before outreach begins.
The app runs as a static page on GitHub Pages. Its interface lets a recruiter enter role context, inspect the generated map, and refine the inputs for another pass.
What you can inspect
- The public app shows the role inputs, generated landscape, and connections between pools and titles.
- The source repository and README show how the map is generated and how to run it.
- A new map can be generated with different role context or exclusions, then checked against market evidence.