Ashby ATS Parser Architecture
How Ashby applies spatial transformer visual block clustering, subword token extraction, and continuous skill-tenure resolution to inbound candidate profiles.
1. Spatial Visual Block Clustering Pipeline
Unlike legacy regex-based ATS engines, Ashby utilizes visual spatial clustering to map PDF bounding boxes into semantic zones before executing sequence tokenization.
+------------------------------------------------------------------------------------+
| ASHBY PARSER ARCHITECTURE |
+------------------------------------------------------------------------------------+
| Inbound Resume Document (PDF / DOCX) |
| | |
| v |
| High-Fidelity PDF Spatial Extraction & Font Table Mapping |
| | |
| v |
| Visual Block Clustering & Layout Segmentation (Transformer Vision Boundary Sweeper)|
| | |
| v |
| Multi-Task Token Extraction & Qualification Node Linking |
| | |
| v |
| Normalized Candidate Graph & Continuous Tenure Model |
+------------------------------------------------------------------------------------+1.1 Layout Segmentation & Coordinate Grouping
Ashby’s pipeline clusters contiguous text coordinates into discrete semantic groups. When resumes use complex multi-column layout tables, the spatial classifier can fragment sentence structures across adjacent columns. Single-column layouts preserve linear reading orders without coordinate collision.
Zero metric hallucination. 100% mathematical token alignment for senior technical & executive hiring.
2. Continuous Skill-Tenure Calculation
Ashby calculates continuous domain tenure by binding technical skills directly to validated employment timeline ranges:
+-----------------------------------------------------------------------------------+
| QUALIFICATION MATCHING ENGINE |
+-----------------------------------------------------------------------------------+
| Requisition Criteria: |
| * Minimum 5 years in Distributed Systems (Go / Rust) |
| * Demonstrated production Kafka or Event Streaming experience |
| |
| Parser Candidate Graph Resolution: |
| * Role 1 (2021-Present): Principal Engineer -> 3.5 yrs -> Kafka, Go -> MATCH |
| * Role 2 (2018-2021): Senior Engineer -> 3.0 yrs -> Distributed DBs -> MATCH |
| |
| Aggregate Qualification State: PASSED (6.5 yrs cumulative verified tenure) |
+-----------------------------------------------------------------------------------+Skills isolated in standalone keyword matrices without direct integration into bullet points fail to accumulate verified calendar tenure inside Ashby’s candidate graph model.
See how Ashby parses continuous distributed systems tenure in our Staff Software Engineer Alignment Guide →
Audit your document against Ashby's Spatial Parser
Verify that your technical stack is anchored within chronological bullet structures rather than isolated keyword blocks in the Axiom Workstation.