Research layer
Not a black box
The matching engine is a research program. Here is the work it builds on, how the four scores are constructed, and where we intend to publish.
Work we build on
- Fairness in ranking and recommendation
- Explainable AI for high-stakes decisions
- Bias auditing in hiring pipelines (FINDHR and related EU work)
- Graph representation learning for careers
How the four scores are built
- Skills alignment
- Shipped a payments SDK and two internal design systems. Matches 9 of 11 must-haves; gap is Kubernetes, which the JD marks nice-to-have.
- Seniority fit
- Led a team of 5 for 3 years. Role asks for 4+ reports. Scoped down from staff to senior-plus after reading the leadership expectations.
- Working-style fit
- Prefers async-first, written decisions, low-meeting weeks. Your team runs the same way. Flags: wants quarterly in-person, you are fully remote.
- Growth trajectory
- Moved IC to lead to IC-by-choice. Explicitly wants deeper platform work, which is the direction of this role over the next 18 months.
Every score ships with a plain-language explanation and a link to the evidence. The candidate sees the same panel the hiring manager does.
Where we take it
- FAccT
- RecSys
- the EU AI Act technical standards work