Enterprise AI Built to Last.

Designed and built the internal equivalent of a general-purpose AI assistant — open-source end to end, connected to real-time business intelligence, operational data, and the full company stack. Developers, business users, and executives reach the same secure system through role-specific interfaces.

The work centers on the unification problem: dozens of systems, incompatible access patterns, no shared context. The platform collapses that surface into one, without surrendering control or data residency to a vendor.

• OpenShift
• Rust
• Neo4j

Built a multi-agent validation pipeline that ingests training content, decomposes it into individual factual claims, and verifies each one against the authoritative source corpus. RAG retrieval grounds every check; specialised agents handle terminology, version drift, and policy alignment.

Self-hosted in the client environment. Reviewers see the unverified claims, not the verified ones — turning a manual QA bottleneck into a focused exception queue.

• Rust
• AxumAxum
• Docker

Built a unified context layer reconciling traces, observability metrics, unstructured knowledge, and IT/data across the entire stack. Agents run on top of it to handle complete production pipelines — code generation, CI/CD, and site reliability — deployed inside the client's perimeter under their governance terms.

Engineers describe what they want; the system generates, tests, and proposes deployments. Humans approve. Nothing ships without sandboxed validation and a named owner.

• EKSEKS
• Rust
• React
• MUI