Deterministic, White-box, Reproducible Hiring Audit Engine.
Axiom-Recruit is a senior-level logic engine designed to evaluate hiring decisions through the lens of mathematical excellence, market alignment, and the "Grit Factor." It eliminates black-box bias by providing a full, auditable logic trace for every hiring verdict. The root candidate.json is the default Ziad Salah profile used by the project.
- Grit Factor (Resource Ratio): Automatically boosts candidates who achieve high technical excellence in low-resource environments (e.g., Egypt vs. Silicon Valley).
- Multi-Layer Logic: Combines Domain Scoring, Derived Heuristics, and Deterministic Constraint checking.
- Deterministic Trace: Every decision generates a
verdict.jsoncontaining the full math, proofs, and a white-box neural correction trace. - Dynamic Market Scenarios: Model decisions across different countries (Egypt, USA, Canada, etc.) and technology eras (AI-Current, ML-2016).
- AI Extraction Prompts: Built-in utilities to extract deterministic feature vectors from unstructured CVs and company descriptions.
git clone https://github.com/Zierax/Axiom-Recruit
cd Axiom-Recruit
# No heavy dependencies; standard Python 3.10+Evaluate a candidate against a company and market:
python axiom-recruit.py --candidate candidate.json --company profiles/companies/coveo/real_world_profile.json --market profiles/markets/canada/ai-current.jsonRun with baseline weights only (ignore saved feedback adjustments in weight_ledger.json):
python axiom-recruit.py --candidate candidate.json --company profiles/companies/coveo/real_world_profile.json --market profiles/markets/canada/ai-current.json --ignore-weight-ledgerEvaluate your own readiness against a specific market using a neutral company baseline:
python axiom-recruit.py --candidate candidate.json --market profiles/markets/egypt/ai-current.json --self-evaluation-modeInitialize or update the /profiles folder with synthetic levels and real-world company data:
python axiom-recruit.py --generate-profiles-libraryweight_ledger.json stores feedback-driven weight adjustments and history.
This means scores can change over time for the same input if calibration has been applied.
- Adaptive mode (default):
--weight-ledger weight_ledger.json - Baseline deterministic mode:
--ignore-weight-ledger - Roll back last calibration event:
--revert-last-feedback
Detailed documentation is available in the /docs folder:
- Engine Logic & Math - How the scoring and Grit Factor work.
- Logical Path - End-to-end decision flow from input to verdict.
- Profiles & Markets - Details on synthetic vs. real-world data.
Axiom-Recruit is a white-box system. It does not "predict" performance; it audits feature sets against explicit corporate and market standards. All weights are adjustable and transparent through the catalog.py and feedback.py systems.
Founder & Lead Researcher: Ziad Salah (Axiom Logic / Division-36) License: MIT