Resources

How matching works

PhdFit does not return a single opaque fit score. It ranks every candidate across five axes so you can see where the match is strong, where it's soft, and where to push back.

The short version

  1. 1. Signals from your CV. The Candidate Analyst reads your résumé and extracts research interests, methods, publications, and trajectory.
  2. 2. Faculty profile corpus. The Professor Researcher indexes tens of thousands of faculty pages, recent papers, and recruiting signals.
  3. 3. Five-axis scoring. Each candidate is scored on topic, method, trajectory, open seats, and certainty — not one number, five.
  4. 4. Explained ranking. The Match Explainer writes the reasoning chain for every match so you can argue with it in chat and re-rank.

The five axes

01

Topic fit

What it answers: Does the professor work on the problems you want to work on?

Signals used: Paper abstracts and keywords for the last three years. Matched against the interests the Candidate Analyst parsed from your résumé plus anything you tell it directly in chat.

02

Method fit

What it answers: Do you share tools, frameworks, and ways of framing problems?

Signals used: Method language in the professor's papers — probes, RL, causal inference, simulation, RCT, etc. — compared with the methods you've actually used, not just listed.

03

Trajectory

What it answers: Is the lab heading somewhere you want to go?

Signals used: Year-over-year change in topics, where recent graduates landed, whether the PI has moved into a new area. Predicts where the lab will be in 3–5 years, not where it was.

04

Open seats

What it answers: Is the professor actually accepting students this cycle?

Signals used: Public lab-page statements, last year's intake size, graduation pipeline, and any "taking students" or "not taking students" language on the faculty page. Shown on every match card.

05

Certainty

What it answers: How confident is the ranking itself?

Signals used: Sample size (how many papers we could analyse), recency (how fresh the data is), and cross-signal agreement. Low certainty does not mean bad fit — it means do more homework before applying.

Presets and when to switch them

The same five axes can be weighted differently depending on what you care about. Balanced is the default — even weights across axes. Switch to Research-fit when you know your exact area and care most about topic and method overlap; Methods-matchwhen you'd rather learn a new topic with tools you already own; Practical-fitwhen you're optimising for open seats and recruitment likelihood. Each preset shows on the match cards in the workspace — you can toggle between them and watch the ranking re-order.

See it on your CV

The easiest way to understand matching is to run it on yourself. Upload a résumé and the first ranked set lands in under a minute.

Open your workspace