The matching pipeline turns two streams of data — funding opportunities and faculty profiles — into ranked, explainable matches. Here's the end-to-end flow.

1. Ingest

• Opportunities are pulled from sources such as Grants.gov (apply07 REST API) on a nightly schedule, with incremental updates throughout the day.
• Faculty come in through institutional rosters and Clubhouse staging, where they can be reviewed before promotion to live data.

2. Enrich profiles

• ORCID is the highest-trust source for publications and is preferred when present.
• Uploaded CVs are parsed by AI to extract expertise terms and prior funding.
• Manual edits by the faculty member are protected — re-enrichment never overwrites them.

3. Match

The matching engine scores each (faculty, opportunity) pair using a weighted blend of topical overlap, publication and funding history, and eligibility signals. The score is shown on every match as a single fit value — see How fit scores are calculated for what feeds it and how to read it.

Each match also carries qualitative reasons — short, icon-mapped tags that explain what aligned (topic, eligibility, prior work) and what didn't.

4. Distribute

• Faculty see matches in Fairway, sorted by fit score by default.
• RD staff see signaled interest and help requests in Signal.
• Leadership sees aggregate engagement and pipeline metrics in Drive.

5. Convert and learn

When a faculty member signals interest, marks Not a Fit, or communicates with RD, the system records the disposition. Those signals feed back into algorithm tuning: weights are reviewed periodically, and faculty corrections take precedence over future re-enrichment of their own profile.

Refresh cadence

• Opportunity ingestion: nightly with incremental updates.
• Profile enrichment: triggered by upload, on-demand, and on a periodic schedule.
• Match recomputation: after either side changes; UI invalidates and re-renders without a manual reload.