Building a local Vision-LLM pipeline that turns streamed YouTube uploads into trustworthy metadata without manually scrubbing through every video.

Refactoring OpenClaw setup, optimizing agents, and integrating local LLMs for efficient automation workflows.

Five weeks building a hybrid OpenClaw setup: hardware, local AI services, cloud models, guardrails, skills, and a seven-agent operating model.

CI/CD cleanup with GitHub organization restructuring, runner migration, monitoring rewrites, and pipeline hardening.

Pipeline cleanup, workflow hardening, and a pragmatic DevSecOps setup focused on minimal overhead and effective secret scanning.

Insights from achieving GitLab CI/CD Associate and iSAQB Foundation certifications—covering pipelines, software architecture views, and stakeholder-driven design.

Adding an Intel Arc A380 alongside my RTX 5080 for dedicated Twitch encoding resolved NVENC crashes and allowed me to deliver smooth high-FPS streams.

Implementing JSON schema and UI testing pipelines with Zod and Playwright to prevent production-breaking changes and enforcing branch protection.

Submitted WTF build for Steam Store review, including .vdf packaging and split redistributables. Automated forum updates via GitHub Actions Windows Runner. Added change detection to avoid full rebuilds. Cleanup included CSS fixes, logic optimizations, and pipeline tweaks.

Pushed the WTF Minimal Model live at minimal.wtf-model.com with 50 descriptions and no overlays. Set up forum previews, Steam deployment via WPF/.NET 8, and CI/CD with Azure ACI. Public repo, intro video, and PDF planned for 2025. Full release delayed but progressing well.