DS
Diwesh Saxena
3× faster shortlist

Talent graph: 3× faster consultant discovery in closed networks

Chief Technology OfficerNoble House ConsultingJan 2024 — Jun 2024Published 14 March 2024

Replaced linear talent-pool search with a graph-based discoverability model for closed consultant ecosystems — context, skills, and relationship-aware matching.

Technology Stack

Neo4jPythonReactPostgreSQLGraphQL

Key Outcomes

  • 3× faster time-to-shortlist for niche consultant requests
  • Graph queries outperforming legacy SQL search on complex skill combos
  • Published as peer-reviewed Zenodo paper on talent graphs
Outcome: 3× faster shortlist for complex consultant searches.

Context

Noble House operates a closed consultant network. Traditional “talent pool” search broke down when clients needed rare skill combinations and relationship context.

Problem

  • SQL LIKE queries missed adjacency (who worked with whom, on what).
  • Recruiters spent hours manually cross-referencing spreadsheets.
  • Privacy constraints prevented open marketplace models.

Approach

  • Model consultants, skills, engagements, and referrals as a property graph.
  • GraphQL API for recruiter UI; hybrid search (vector + graph traversal).
  • Privacy rules encoded at query layer — no raw PII in graph exports.

Architecture

PostgreSQL (source of truth) → sync jobs → Neo4j → GraphQL → React recruiter UI.

Research

Full framework published: From Talent Pools to Talent Graphs.

Results

Median time-to-shortlist dropped by two-thirds on multi-skill requests; recruiter satisfaction scores rose in Q2 2024 pilot.

CTA

Building discoverability in closed talent networks? Book a 30-min call.