Anatomy of a Job Fit Score
Exactly how our hybrid deterministic + LLM score is calculated — with weights.
Most resume scoring tools are black boxes. Ours isn't. Here's the full recipe.
The weights
| Component | Weight | Source | |------------|-------:|-------------------------------------------| | Keywords | 30 % | TF-weighted overlap (deterministic) | | ATS | 15 % | Rules engine (deterministic) | | Experience | 15 % | Date-range + word-count heuristic | | Semantic | 25 % | LLM semantic fit | | Seniority | 15 % | LLM seniority vs JD level |
Why 60 / 40?
Because AI scoring alone is unreliable — small prompt tweaks can swing scores by 20 points. Pinning 60% of the score to deterministic signals caps drift.
Why one LLM on OpenRouter?
We route LLM tasks through a single OpenRouter-hosted model — one fast, cost-aware stack for scoring and rewrite, without cross-model fallback.
What this means for you
Two resumes with the same keyword overlap can still diverge by 10 points if one communicates seniority more clearly. The free tool shows you which bucket is dragging you down.