AI-/Cloud-Transformation Consultant · Senior IT-Projektmanager · Open-Source-Maintainer
I help teams turn complex technology programs into working systems: AI automation, cloud platforms, data products, digital platform delivery, and reliable IT operations.
Public-safe proof points include work around BMW, Porsche Bank, and Volkswagen projects across AI, analytics, regulated platform delivery, connected services, and operational enablement.
| Area | What I build and improve |
|---|---|
| AI automation | Coding agents, maintainer workflows, triage helpers, prompt/tool log safety, and practical copilots |
| Cloud platforms | Delivery models, platform foundations, operational enablement, and scalable cloud adoption |
| Data products | KPI foundations, analytics workflows, reporting systems, and decision-support tooling |
| Digital programs | Roadmaps, stakeholder alignment, delivery governance, vendor coordination, and launch execution |
| Reliability | Security-aware processes, review loops, release readiness, and maintainable operations |
I maintain small, practical tools for developers and open-source maintainers. The current focus is agent infrastructure, MCP diagnostics, repository readiness, and privacy-aware workflow automation.
| Project | Purpose |
|---|---|
| mcp-server-doctor | Validate JSON MCP capability reports from files or stdin |
| mcp-permission-policy-tester | Scan MCP tool descriptions and schemas for risky permission policy signals |
| prompt-tool-log-redactor | Redact common secrets and PII from plain text and JSONL logs |
| oss-repo-launch-doctor | Check open-source repository launch readiness |
| context-budgeter | Rank repository files against a task query and token budget |
| coding-agent-eval-harness | Run deterministic fixture-based evaluations for coding-agent tasks |
| maintainer-inbox-triage-cli | Dry-run local triage for GitHub issue and pull-request JSON payloads |
| agent-trace-lite | Record simple agent spans/events to JSONL and summarize them |
- Building practical OSS tools around Codex, MCP, and AI-assisted maintenance
- Improving repository readiness, contributor onboarding, and release hygiene
- Turning agent workflows into deterministic, testable command-line utilities
- Keeping security, privacy, and operational usefulness visible from the first commit
- Clear scope before implementation
- Fast prototypes, but with tests and maintainable structure
- Pragmatic documentation that helps contributors act quickly
- Security and privacy checks before publishing
- Delivery communication that stays concrete and decision-oriented
- Website: sebastian-albrecht.de
- Open source hub: sebastian-albrecht.de/open-source
- Profile: sebastian-albrecht.de/sebastian-albrecht
- German profile: sebastian-albrecht.de/de/sebastian-albrecht
- Cases: sebastian-albrecht.de/cases.md
- AI automation: sebastian-albrecht.de/services/ai-automation
- Cloud & DevOps: sebastian-albrecht.de/services/cloud-devops
- Data analytics: sebastian-albrecht.de/services/data-analytics
kontakt@sebastian-albrecht.de