Lukas Grigis

• AI-augmented software engineering

  Agentic workflows and Specification-Driven Development in client teams: the context, instructions, skills and guardrails that make coding agents reliable, and multi-agent workflows where small models implement and strong models gate.

  • Built and published ralphctl, a harness for long-running AI coding tasks across Claude Code, Copilot CLI, Codex, OpenCode and Grok.ralphctl on GitHub (opens in a new tab)ralphctl on npm (opens in a new tab)
  • Built the AI Coding Sandbox, a containerized, policy-controlled environment for running coding agents safely.
  • Scaled an SDD framework across a large enterprise's central AI-in-software-development program.
  • Part of Zühlke's AI-augmented delivery initiative: methodology, tooling and coaching across projects.

• AI engineering

  GenAI features end to end: RAG from naive to agentic, LLM evaluation, structured output, tool use, MCP servers and clients, local-first models.

  • Technical lead on a RAG-as-a-service platform where small and medium businesses build their own AI assistants.
  • Built context-vault, an LLM-maintained knowledge base with citations, local on Ollama and callable over MCP.
  • Built spring-oauth2-mcp, an OAuth2-secured MCP server with Dynamic Client Registration in front of a federated identity provider.spring-oauth2-mcp on GitHub (opens in a new tab)

• AI governance and safe adoption

  AI adoption an organization can audit and defend: guidelines, use-case approval, risk assessment, rules for AI tools inside client environments.

  • AI lead on a large public-sector digital transformation: authored the binding AI guideline and built the governance register for AI use cases, services and MCP servers.
  • Defined practical guidelines for using AI tools safely inside client environments.

• Teaching and enablement

  Courses and workshops that make teams and leaders fluent in AI, with material that runs.

  • Created and teach the AI Engineering course: naive RAG by hand, advanced and agentic RAG, evaluation and MCP, in Java and Spring AI, fully local.
  • Teach AI-augmented software engineering courses at Zühlke and for clients.
  • Ran a hands-on AI workshop where senior management built a governance-tool prototype in five hours.

• Engineering foundation

  Identity and security (OAuth2/OIDC, Keycloak, RFC 8693 token exchange, broker security), distributed and event-driven systems, cloud-native platforms on AWS and Azure.

  • Reference implementation of RFC 8693 token exchange between gateway and services, with Keycloak.spring-oauth2-token-exchange-rfc8693 on GitHub (opens in a new tab)
  • Securing broker connections and per-message permissions in RabbitMQ with OAuth2 (AMQP 0.9.1 and 1.0).spring-oauth2-amqp-0.9.1 on GitHub (opens in a new tab)spring-oauth2-amqp-1.0 on GitHub (opens in a new tab)

ExpertAdvancedProficientFamiliar