Services for teams building AI workflows that need to keep working.

I help with systems where the hard part is not the first model call, but the surrounding workflow: source changes, review loops, reliability, retries, documentation, and operational ownership.

Focus

  • Production LLM systems

    Architecture review and implementation support for RAG, structured outputs, model fallback, evaluation, safety filters, and operational failure modes.

  • GitHub-native AI automation

    Designing automation that fits issues, pull requests, CI/CD, code review, release constraints, and maintainer control instead of bypassing them.

  • AI localization infrastructure

    Markdown-aware translation pipelines, synchronized multilingual documentation, versioned translation assets, and reviewable PR workflows.

  • Developer education

    Technical writing, demos, workshops, and explainers for Azure AI, RAG, fine-tuning, Responsible AI, Semantic Kernel, and production LLM workflows.

Contact

  • Start with a practical problem

    Send a short note to minseok.song@mssong.com if the work involves LLM systems, developer automation, localization infrastructure, or public technical education.