Facultas. Request access

Matching infrastructure

Job matching that knows who actually gets hired.

Facultas is the backend matching engine for platforms that connect students to internships and jobs — ranked, explained recommendations grounded in fifteen years of real placement outcomes, not a generic job board.

Request access Read the approach

fa·cul·tas  ·  Latin  ·  capability, the means, opportunity

01 The thesis

Most job boards rank by keywords and recency. They have no memory of which organizations actually hire from a given population. We do.

Facultas maintains a curated, continuously refreshed database of opportunities and matches each profile with hybrid semantic and keyword retrieval. It then ranks with a placement prior — a signal learned from fifteen years of real outcomes — so proven employers surface ahead of plausible-looking noise.

Every recommendation comes back with a per-criterion breakdown and a link to its original posting. No black box, no dead ends.

02 What makes it different

A placement prior, not a guess.

Fifteen years of real placements — 1,395 across 759 employers — tell us which organizations actually hire this population. Proven targets rank higher; the long tail doesn’t drown them out.

Hybrid retrieval.

Semantic search finds meaning; full-text search keeps exact signals — FPGA, OPT, Series 7, specific agency names — from being lost. We fuse both, so neither precision nor nuance is sacrificed.

Explained and traceable.

Every match returns a per-criterion score breakdown and a link back to the source posting. Hard constraints — work authorization, dates, sector exclusions — are gates, never suggestions.

  • 15years of placement history
    2011–2025
  • 1,395real placements
  • 759unique employers
  • 71countries

De-identified historical placement data — the ranking signal no public job board has.

03 The pipeline

  1. 01

    Curate

    Sources are chosen from proven employers first — official agency feeds, public ATS boards, and vetted aggregators.

  2. 02

    Normalize

    Each posting is structured into one clean schema: skills, sector, location, visa flags, deadlines, seniority.

  3. 03

    Embed

    Postings are embedded and indexed for fast semantic and keyword retrieval — kept fresh on a recurring cycle.

  4. 04

    Match

    A profile arrives; hard constraints filter, hybrid retrieval ranks, and explained matches return in milliseconds.

04 Built to integrate

A documented REST API. Send a profile, receive ranked matches.

API-key authentication, per-client rate limits, and a clean path from prototype to cloud — the architecture that runs the demo is the one that scales. The end-user experience stays yours; the matching is ours.

POST /v1/match
profile     field: International Affairs
            skills: [Python, GIS]
            needs_sponsorship: false

→ matches
   World Resources Institute   score 0.87
   why: climate-policy fit · GIS + Python
        required · 6 prior placements

Want it powering your platform?

We’re partnering with a small number of opportunity platforms. Tell us about yours.

Request access