Three days at the CGI AI Hackathon during Silicon Saxony Days 2026: building an AI prototype with AWS, Terraform, Dify.ai, Amazon Bedrock, Titan embeddings, and RAG.

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.

Efficiency improvements through faster compression, reduced complexity, and aggressive media optimization to significantly lower deployment size.

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.

I replaced a costly VM-based Discourse forum with a zero-cost Azure Static Web App powered by Hugo and OpenAI Codex. I detail the rationale, savings, automated workflow, and features added—dark mode, client-side search, responsive design, and newsletter integration.

Launched a Discourse-based forum at forum.wtf-model.com for WTF Model discussions. Used Bitnami image on Azure and a custom .NET8 app to auto-post entries via API. Handled image merging, metadata, and rate limits. OAuth 2.0 login enabled; all content now accessible.

Migrated WTF project from Azure VM to Static Web Apps for cost and scalability. Replaced WooCommerce with Shopify and refactored backend logic into Azure Functions using C#. Integrated key generation, storage, and GitHub Actions deployment. Total cost now under €25/month.