Methodology

career-ops-ui fronts the career-ops evaluation engine: instead of amplifying applications, it filters them. Every job listing is scored against your CV on a six-dimension 0.0–5.0 rubric, and the score decides the next step — the rubric is published, every evaluation cites your own materials, and nothing is ever auto-submitted.

The score thresholds

A score below 4.0 is a recommendation not to apply. That is the point of the system: ten applications you believe in beat two hundred you don't.

ScoreNext stepCLI
≥ 4.5High fit — apply immediately/career-ops apply
4.0 – 4.4Apply, or seek a warm intro first/career-ops contacto
3.5 – 3.9Research the company first/career-ops deep
< 3.5Skip unless you have a specific reason

The six dimensions

Each evaluation scores the role match, alignment with your long-term goal (the north star), compensation against your floor, cultural signals, red flags in the posting, and global fit (location, visa, language). The verdict is LLM judgement against a published rubric — not closed-form math — so every score comes with cited evidence from the posting and your CV.

What it never does

No auto-apply — a human sends every application. No invented facts — every CV claim traces back to your own materials. No cloud accounts, no telemetry, no selling your data: everything runs locally on your machine.

These rules are the CareerOps Manifesto, applied: career-ops.org/manifesto

Full methodology — career-ops.org/methodology

Developer docs & guides — career-ops.org/docs