I work on SAP operations, enterprise architecture, integration, transformation assurance, data, and practical AI. My public repositories are designed as bounded tools and contracts: each project should help with one decision, produce an inspectable artifact, and state what it does not prove.
Open the problem-led project map · Read the knowledge base · Discuss a concrete problem
| If you need to… | Start here | What you should get |
|---|---|---|
| Turn enterprise context and constraints into an explainable design | Enterprise Architecture Composer | An architecture proposal with alternatives, constraint evaluation, and a decision trace |
| Render structured business or architecture material as a deterministic visual | Visual Workbench | A reviewable SVG projection whose imported semantics remain owned upstream |
| Connect project claims to artifacts, tests, freshness, and review status | Project Evidence Graph | An assurance graph and evidence pack with explicit missing-evidence findings |
| Explore bounded agent patterns for SAP operations | SAP Agentic Operations | Safety, evaluation, and operating patterns without live SAP execution authority |
| Turn public sources into provenance-aware research context | Signal to Insight | Reviewable source, insight, and handoff artifacts without automatic operational authority |
The public project map covers the supporting contracts, graphs, controls, datasets, and agent-facing profile tools without presenting every repository as an equal starting point.
The Enterprise Change Evidence Pack follows a synthetic change from source-backed context through an architecture decision, a business-readable visual, and an assurance graph. Every handoff is labelled as implemented, documented, or demonstration-only; a digest or logical reference is never treated as approval.
Run the reference case · Inspect the machine-readable portfolio
- Start from the decision or operational problem, not the technology label.
- Keep source ownership and derived projections separate.
- Prefer deterministic fixtures, validation commands, and inspectable outputs.
- Use synthetic or redacted public examples; keep enterprise data private by default.
- Treat human review, business approval, and execution authority as explicit boundaries.
Created and maintained by Dzmitryi Kharlanau, an SAP consultant and system analyst working across enterprise architecture, data, integration, operations, and practical AI.