back to work
AI Applicant Tracking
A recruiting pipeline with an AI evaluation layer that scores interview responses against a rubric — so the team opens a ranked shortlist instead of wading through raw transcripts.
The problem
Every applicant produced a wall of free-text interview answers. Reviewing them fairly meant reading each one end to end, holding the rubric in your head, and trying to stay consistent across dozens of candidates — slow, and inconsistent by the afternoon.
What I built
An evaluation layer that sits on top of the existing applicant data and does the first read for you.
- Rubric-based scoring — an LLM grades each response against the same explicit criteria a human would, every time.
- Ranked output — scores roll up so the team sees candidates ordered, not a flat list of transcripts.
- Human-in-the-loop — the AI ranks and explains; people still make the call, just on a curated shortlist.
- Airtable-native — it lives where the recruiting data already lives, no new tool to learn.
Outcome
The team stopped reading raw transcripts to find the top few. Scoring is consistent across every candidate, and review time goes to the people most likely to be a fit.
Rankednot raw transcripts
1consistent rubric
In useby the team
// proof
A look inside
