The challenge
Candidate information can be spread across spreadsheets, Drive folders, CVs, databases, and Workable. Technical interviews may also appear as internal events, so filtering by event type alone misses relevant records.
The workflow
Candidate sources in Sheets and Drive
Documented CV selection and identity rules
Traceable PostgreSQL imports
Candidate, job, and event data in Workable
A closer look
A candidate folder is matched to source information, a CV is selected by documented rules, and provenance is stored in PostgreSQL. A repeated import updates the same record; an operator reviews before a Workable write.
Evidence & status
- Authenticated access to the Workable API was verified.
- Six technical-interview title matches in an initial sample of 20 events.
- The matches appeared as InternalEvent in Workable.
- Idempotent imports update existing records instead of creating duplicates.
Local TypeScript toolkit. The 20-event sample is not a complete interview inventory. The source document does not describe a specific AI inference capability, so this story focuses on integration and data-quality engineering.
An engineering decision
An event participant is not necessarily the technical interviewer, and an accepted RSVP is not proof of attendance. The workflow preserves these distinctions.
The opportunity
The opportunity: more consistent records, fewer duplicate imports, and easier review. Production migration-time savings and reporting completeness have not been measured.
Adapted from “NAB Oct 2026 — AI Use Cases & Stories.” Statuses and findings reflect that source. Names are listed as project contacts, not credited authors.