Skip to content
Rachit Gupta

Projects · 01

Ontology-Grounded Agentic AI Data Steward

An engineering exploration of how deterministic entity-resolution evidence, governed enterprise context, bounded AI recommendations and human-review boundaries can work together — preserving clear distinctions between evidence, uncertainty, review and authority.

Independent engineering projectStatus: active prototype, 2026–present

Problem

Reliable enterprise context is necessary for trustworthy AI, but it is not sufficient on its own.

Agents that act on enterprise data inherit every ambiguity in that data — duplicated identities, unclear ownership, missing lineage. Before an agent can be useful, it needs context it can rely on, and boundaries that make clear what it knows, what it is guessing, and what it is allowed to do.

Current exploration

  • Identity resolution
  • Structured evidence
  • Governed agent capabilities
  • Human-in-the-loop stewardship
  • Enterprise context and governance

Engineering philosophy

  • Facts should have authoritative sources.
  • AI recommendations should not silently become system actions.
  • Missing evidence should remain missing evidence.
  • Human decisions and machine authority should remain distinguishable.

Current limitation

The prototype produces recommendations for review. It has no authority to modify enterprise data, and it is not a deployed or production-validated system.

Selected enterprise work

Professional experience, not personally owned projects.

Examples of product work delivered for enterprise clients. No client code, diagrams or confidential material is shown.

01

Enterprise Data Marketplace

Product delivery for enterprise data discovery, metadata, governed access and approval workflows.

02

Master Data Management and Customer 360

Functional product work covering data matching, duplicate resolution, stewardship and governed master data.

03

ML-Driven Pricing and Decision Support

Product experience connecting merchandising requirements with predictive analytics and optimization workflows.

04

Enterprise Payment Platforms

Requirements, integrations and modernization work for enterprise payment processing and client implementation.

Working on similar problems? I'd be glad to compare notes.