Projects
Semantic model validation, structured PBIP/PBIR engineering, agent harness patterns, and public-good decision workflows each isolate a specific discipline so the approach is visible independent of any single client engagement.
Power BI Reference
These static reference pages show the decision framing, public source gate, model plan, validation notes, and publish-readiness path before any Power BI iframe is allowed.
Static reference
A public-data Power BI reference demo scaffold showing how airline reliability data can move from decision context to source-aware actions, with semantic model proof and validation gates visible before any iframe is published.
Repository Gallery
Open repositories stay visible first so technical proof is immediately inspectable, including recent applied-AI builder work from OpenAI-related events.
powerbi_demo_PBIPxGHCopilot
Automated DAX measure testing built on PBIP, Python, and AI-assisted tooling, designed to catch calculation risk before deployment.
Validation-first BI engineering: systematic DAX risk detection, repeatable test coverage, and CI-ready semantic model workflows.
modellismz/OpenSkillTrace
Sea x OpenAI Codex Hackathon build: an AgentOps harness for tracing agent runs, routing fallbacks, checking eval and policy signals, and keeping approval and audit controls visible.
Forward-deployed AI is not just prompt output. The useful work is the control layer around tools, data, fallbacks, and human decision gates.
disaster-dashboard-webapp-repo
Disaster-response dashboard copilot shaped around the Asia Pacific AI Skills Jam: profile fragmented CSV and Excel files, surface evidence gaps, recommend dashboard views, and export a reviewable decision handoff.
Applied AI with BI discipline: start from messy operational data, keep deterministic checks visible, use AI only where it improves reasoning, and leave decision ownership with humans.
All public repositories focus on reusable methods and engineering patterns. Client-specific implementations remain private; what you see here reflects the discipline applied across every engagement.
Internal Patterns
These stay visible as evidence of working methods, but not as the first proof block.
PBI_Agent
A deterministic PBIP and PBIR engineering template that enforces validation gates, consistent folder structure, and governance-ready defaults.
Structured engineering habits that transfer across projects: repeatable repository design, clear validation checkpoints, and governance-aware BI delivery.
A2A
A multi-agent framework exploring AI-assisted orchestration across Azure, Power BI, and Databricks BI workflows.
Early-stage but grounded: tests whether multi-agent patterns can reduce manual coordination in real BI delivery pipelines.