
DevScout ranks engineering candidates on what they have actually built and solved — résumés parsed to their hidden links, competitive-programming profiles pulled live, and every score broken down into sub-scores you can argue with.
Three failure modes stand between a stack of résumés and a shortlist you can defend.
Every candidate is scored on live public performance data, never on self-reported skill lists alone.
A résumé asserts skill. Nothing in the PDF proves the candidate ever shipped or solved anything.
Most profile links sit behind display text like “LeetCode”, invisible to a plain text scrape of the page.
Platforms restyle their markup and a screening pipeline starts returning nulls that look like weak candidates.
The five things DevScout does between a folder of résumés and a shortlist you can defend.
CANDIDATE-RESUME.PDF
Links · LeetCode · Portfolio
The visible text just says “LeetCode”. The handle itself lives in the PDF’s link annotation — which is where DevScout reads it from, not the page text.
Helping teams hire engineers on proof of work — résumé text, hidden profile links, and live platform performance resolved into one explainable score.

We turn a folder of résumés into a ranked shortlist through one repeatable, auditable pipeline.

Résumé PDFs are parsed for both text and embedded link annotations, then LeetCode, Codeforces, CodeChef and LinkedIn handles are auto-resolved for the recruiter to correct.
Bright Data custom scrapers pull live public performance data in a bounded async fan-out, and every candidate comes back with sub-scores and a confidence badge.
Frontend, backend, cloud, ML or new grad — one pipeline, re-weighted for what the role actually needs.
SCORING WEIGHTS
Frontend Engineer
Weight profile { frontend }