Case study — research security

AURA: AI risk analysis
for research security

A research-security consultancy hired KVEC to build a platform that maps researchers' publications, funding, and affiliations into a knowledge graph and generates AI-driven risk reports. It's in active use for government research-security work today.

federally funded researchers mapped into the knowledge graph
21,000+
federally funded researchers mapped into the knowledge graph
users the client rolled the platform out to
60+
users the client rolled the platform out to
typical bug-fix turnaround, not sprints
Hours
typical bug-fix turnaround, not sprints

The challenge

Manual review doesn't scale to 21,000 researchers.

Research-security analysts assess whether federally funded researchers have undisclosed foreign affiliations, funding, or conflicts. The raw material is public — publications, grants, institutional records — but it's scattered across dozens of sources, and reviewing one researcher by hand can take hours.

The client needed that review to happen across an entire federally funded research portfolio — consistently, defensibly, and fast enough to be useful to government clients.

What we built

One platform, end to end.

DATA

Ingestion pipelines

Automated pipelines pulling researchers' publications, funding records, and institutional affiliations from public and licensed sources into one consistent model.

GRAPH

Knowledge graph

Researchers, institutions, funders, and co-authorship networks connected as a graph — so an analyst can see relationships that never surface in a spreadsheet.

AI

AI-driven risk reports

Structured risk reports generated per researcher, grounded in the graph — turning hours of manual open-source review into minutes of reading.

SEC

Security & access controls

Role-based access, audit-friendly logging, and the security posture government research-security work demands — including client security questionnaires.

How we worked

Like a team member, not a vendor.

  • Founder-led: the engineer on every call was the engineer writing the code
  • Features shipped continuously — including ones the client didn't know to ask for
  • Bugs almost always fixed within hours of discovery
  • KVEC builds and operates the platform today — not a handoff and goodbye
KVEC exceeded all my expectations. Eugene has implemented every feature I requested and added features I did not know I'd want until I saw them. If I ever found a bug, he almost always fixed it within hours of discovery. He has never told me that a feature I wanted was impossible — he always finds a way to make his customer's vision come to life. I've already made the tool available to roughly 60 individuals, and their reviews have been overwhelmingly positive. I'm very glad I hired KVEC, and hope to hire them for future projects.

Founder & CEO

Research-security consulting firm

Need something like this?

Data platforms, knowledge graphs, and AI reporting — built and operated by the engineer you talk to. Government and larger engagements scoped individually.