Scaler HireMaster

Overview

HireMaster turns the problem of hiring from Scaler into a self serve product. An employer posts a job, the entire learner pool is scored and ranked against it, contact details unlock only on shortlist, and the whole hiring lifecycle runs as a tracked pipeline with a live funnel.

The Problem

Scaler has a large pool of vetted, upskilled learners and a steady stream of employers who want to hire them, but matching the two was manual, opaque, and slow, with no consistent notion of fit and no safe way to expose learner data to an outside company.

Key Features

• Post a job with structured requirements: skills (must have and nice to have), CTC band, experience, work mode, location, notice period.
• Every learner passes hard eligibility gates, then survivors get one blended Fitment Score from 0 to 100 and are shown as a single ranked list.
• Contact details stay masked until a candidate is shortlisted, enforced in the database via a view, not just hidden in the UI.
• A full hiring pipeline: Shortlisted, Contest, Interview, Offer, Hired, with a live stage to stage conversion funnel.
• A reach out composer to email candidates directly, with a record of past outreach.

How It Works

1. 01Employer signs in with Google OAuth, restricted to verified company domains.
2. 02Posting a job triggers the matching engine, which filters the entire learner pool through five hard eligibility gates, then scores survivors with a blended Fitment Score.
3. 03Matches render as one ranked list with contact details masked at the database level via a Postgres view.
4. 04Shortlisting a candidate unlocks contact info and enrolls them in the hiring pipeline, from Contest through Interview to Offer.
5. 05A live funnel tracks stage to stage conversion, and Vercel functions run in Singapore to sit next to the database.

Product Thinking

• Single sided by design. Only the employer gets a UI. Learners are data and never log in, which removes an entire class of privacy, moderation, and cold start problems.
• Filter then rank, not score everything. A candidate who fails a hard gate is never surfaced, so the list only ever contains people who could actually be hired.
• The displayed score is a real fit score: 70% skill alignment plus 30% credibility, replacing an earlier two bucket Strong versus Potential split that implied more precision than the data supported.
• Identifiers are secrets too. The learner sheet ID and infra identifiers are treated as secrets, not just passwords, because the underlying pool is real PII.

Tech Stack