Score CVs against your job spec
CV & Talent Intelligence
Parse, extract, and structure CVs with a 6-stage LLM pipeline built for shortlist support. PII guardrails, schema-validated confidence scores, and human review hooks included.
Key Capabilities
6-Stage Extraction Pipeline
Parser -> skills -> experience -> projects -> preferences -> validator. Each module runs with deterministic configuration and few-shot prompts for consistent, reproducible extraction.
PII Guardrails
Built-in PII redaction scans every extracted field. Social security numbers, bank accounts, and personal IDs are stripped before storage.
Schema-Validated Output
Every field passes through a strict typed schema with field extraction, per-field confidence scores, and structured arrays. Port contracts enforce compatible I/O between modules.
Content-Hash Deduplication
CVs are hashed before processing. Duplicate submissions return cached results instantly - zero wasted LLM calls.
Skill Alias Normalization
"React.js", "ReactJS", and "React" all resolve to the same canonical skill via the alias expansion engine.
Batch Processing
Process CVs in batches via provider batch APIs. Automatic fallback to sequential for smaller batches.
How It Works
Upload CVs
PDF, DOCX, or plain text. Single or batch upload via REST API or drag-drop UI.
Pipeline Extracts
6 modules run sequentially with deterministic config. Each produces structured JSON with confidence scores.
Validate & Normalize
Three-layer validation: schema rules -> business rules (email, phone, date) -> skill alias normalization.
Search & Rank
Query candidates by seniority, skills, location, and salary range with fast structured search.
Technical Stack
Ready to build production
Self-host in minutes with Docker, or use the cloud. Either way, you own your data and your models.