Lead Search/ Document Linking Engineer

Lead Search & Document Linking Engineer — Public Records
Public Records Search Portal

All candidates should make sure to read the following job description and information carefully before applying.

Employment type:

Full-time or Contract
Location:

Hybrid/ Remote
Seniority:

Senior / Specialist
Team:

Works alongside our existing full-stack engineering team

About the Role
We operate a public-facing portal where residents, title companies, attorneys, and researchers search and order official county records — deeds, mortgages, liens, plats, court filings, UCC filings, and similar instruments.
We are hiring a specialist for our document linking, or “related records,” capability. The goal: from any single record, automatically surface every related document — same parties, same parcel, referencing or referenced instruments, and full document chains such as deed → mortgage → assignment → release.
This is a focused, deep role. You will own the data modeling, entity resolution, and relationship layer that powers related records. Our full-stack team owns the portal UI and enterprise workspace features and will integrate with what you build, so your core deliverable is clean, linked data exposed through well-documented APIs.

What You'll Do
Build the related-records engine

using entity resolution and record linkage techniques.
Link records across the corpus by parties (grantor/grantee, debtor/creditor), parcels/APNs, instrument numbers, book/page references, and explicit cross-references.
Reconstruct document chains

(e.g., deed → mortgage → assignment → release) so any record returns its full, accurate network of related filings.
Normalize and standardize messy data — person and business names, addresses, and legal descriptions — so matching is reliable across legacy and OCR'd records.
Select and implement the right matching approach: deterministic rules, probabilistic/fuzzy matching, and/or ML-based linkage, with confidence scoring.
Model relationships in a graph database or graph layer (e.g., Neo4j or equivalent) and expose them via clean APIs for the full-stack team.
Build indexing and ETL pipelines from the system of record, handling scanned/OCR'd documents and inconsistent legacy data.
Define and track link-quality metrics

(precision/recall on matches, false-link rate) and tune iteratively.
Apply redaction and PII protection in line with applicable public-records and privacy laws.
Document the data model and matching logic, and train internal staff for ongoing maintenance.

What You'll Bring
Required
Entity resolution / record linkage / MDM:

proven, hands-on production experience.
Data matching:

deterministic and probabilistic/fuzzy matching, blocking/candidate generation, and confidence scoring (familiarity with tools such as Dedupe, Splink, Zingg, Senzing, or comparable).
Graph databases / knowledge graphs:

experience modeling and querying relationships (Neo4j or similar).
Data engineering:

ETL/indexing pipelines, schema design, and large-scale data normalization and cleansing.
Programming:

proficiency in Python (or your stack) working with both structured and unstructured (OCR/document) data.
APIs:

ability to expose results cleanly for a separate front-end/platform team to consume.

Preferred
Experience with public records, land records, title/recording systems, or legal/government data (deeds, liens, UCC, court records).
Familiarity with parcel/APN data, legal-description parsing, and title-chain concepts.
ML-based entity resolution and address
ame standardization.
PII redaction and public-records/privacy compliance.
What Success Looks Like
From any record, users reliably see the complete and accurate set of related documents.
Document chains (deed → mortgage → release, etc.) are correctly reconstructed.
Links are scored and measurable, with a low false-link rate. xsgimln
The system is fast, scalable, and maintainable by internal staff.