Hiresense
A first-round interview with a score a reviewer can check.
Overview
Hiresense runs a live interview, scores communication, confidence, and technical knowledge from 0 to 10, and stores the reasoning plus a quote from the candidate's own words. A person still records advance, hold, or reject. The model does not hire anyone.
The Problem
A first screen is slow when every answer has to be read by hand, and an automated score is useless if nobody can see why it was given.
Key Features
- JWT sign-in for candidates and recruiters
- Live interview over Socket.io, with questions from Gemini
- Scores of 0 to 10 with a reason and an evidence quote
- CV match when a Gemini key is set; PDF only, 10 MB
- Reviewer decision endpoint. The model does not hire anyone
- OpenAPI docs and an eval set with a name-swap check
Technology Stack
My Contribution
Built the candidate and recruiter apps, the scoring agent, and the review step that keeps the hiring decision with a person.
Challenges
Without an OpenAI key the scorer falls back to length, hedging, and a fixed term list. That is explainable. It is not a measure of job performance.
Outcome
A public sample of an interview pipeline where the score can be audited and the decision stays with a reviewer.
Interested in working together?
Tell me about your project and let's see if I'm a good fit.