Founder & Head of R&D at RALFORION. Building OrionBelt®, an open-source Semantic Sidecar for Agentic AI. So AI stops guessing your GROUP BY.
Tip
Currently working on: The OrionBelt® ecosystem — a Semantic Sidecar for Agentic AI with ontology generation, governed text-to-SQL, MCP server, Apache Arrow Flight, and PostgreSQL Wire Protocol. Because data deserves better than copy-pasted SQL.
- Building OrionBelt® Semantic Layer, OrionBelt® Analytics & OrionBelt® Ontology Builder — because data deserves better than copy-pasted SQL
- Semantic data modeling, knowledge graphs, and BI architecture — the stuff that makes dashboards and Agentic AI actually trustworthy
- Longtime Qlik community contributor — extensions, open-source tools, and opinions nobody asked for
- Longtime Gopher 🐹, now addicted to snakes 🐍
- Arctic Code Vault Contributor — my code will survive the apocalypse, no pressure
Ontology-driven analytics platform for AI agents — transforming how data is modeled, explored, and understood through knowledge graphs and semantic reasoning.
A universal semantic layer that bridges the gap between raw data and business meaning, enabling consistent metrics and definitions across tools and teams, designed with an API-first approach for agentic AI.
A Streamlit-based application for building, editing, and managing OWL ontologies — providing a visual interface to define classes, properties, and relationships for semantic data models.
Run OBML query batches against the OrionBelt® Semantic Layer and emit reports. YAML-defined runs produce self-contained markdown, HTML, or PDF reports with audit-grade YAML run-logs. Built for cron, CI, and scheduled audits.
Conversational AI interface tying the OrionBelt® platform together. 300+ models via OpenRouter, Anthropic & OpenAI direct, local LLMs via MLX or Ollama. Dual MCP server support with sampling, inline Plotly charts, and Mermaid diagrams.
X-ray your MCP server — token tax, tool confusion, and surface bloat, distilled into one graded report.
A fast command-line converter from Qlik QVD files to Parquet. Preserves real Parquet types instead of stringifying everything, keeps MONEY and FIX columns as exact decimals, decodes records in parallel, and streams row groups so large files never need to fit in memory.
- What is a Semantic Sidecar? — the pattern OrionBelt® implements (drop-in governed semantics for AI, analytics, and data systems, no architecture change)
- Agentic AI Data Access — governed, consistent, audit-ready access for AI agents via MCP
- Governed Text-to-SQL — fan-trap prevention via ontology, AST, and MCP
- OrionBelt® Platform Overview — the complete stack on one page
