Framework Engineer

Responsibilities

What you should bring

What we offer you

• You will design and develop reusable framework components for an AI platform, enabling other teams to build and operate AI-powered automations.
• You will work with architecture to develop technical concepts and document decisions comprehensibly in Architecture Decision Records, arc42, and C4 diagrams, including discarded alternatives.
• You will be responsible for framework components throughout their entire lifecycle: design, schema and contract, implementation, tests, documentation, versioning, and further development.
• You will develop structures for the framework and runtime that allow other teams to define, execute, and deliver agents and automations.
• You will design schema and contract design for different component types such as Agents, Skills, Tools, Knowledge, Hooks, Contracts, and Guardrails, ensuring versioning and backward compatibility.
• You will build registries and distribution mechanisms so that components can be versioned, registered, found, and integrated via tooling.
• You will develop an extensible plugin and adapter architecture with clear extension points, automatic detection of new components, and high maintainability.
• You will design Guardrails as a technical and functional component, considering authentication, authorization, secrets, isolation, approvals, protection mechanisms, and cost and rate limits.
• You will connect multiple AI providers and chat completion interfaces and develop solutions for routing, model selection, batching, retry, and rate limiting.
• You will extract knowledge from code and make it available as a usable knowledge source in the platform ecosystem via appropriate protocols and mechanisms.
• You will set up CI/CD pipelines in GitLab, anchor quality checks as blocking stages, and establish reproducible release and distribution processes.
• You will anchor Spec-driven Development in the team and ensure that requirements, designs, and test commands are clearly described before implementation.
• You will design and operate the orchestration of multiple agents, including state management, resumption after interruption, loop detection, and parallelization rules.
• You will make the quality of results from agentic processes measurable, define test cases and evaluation metrics, and establish repeat checks for changes and model switches.
• You will build observability for agentic processes, creating transparency regarding runtime, token consumption, costs, and allocation to teams and use cases.
• You will ensure execution isolation, considering sandboxing, minimal privileges, separation of data and access, and the limits of these protection mechanisms.
• As a maintainer, you will support Inner-Source contributions from other teams, review them technically and functionally, and assist with their independent use of the components.
• You will impart knowledge through documentation, demonstrations, office hours, and similar formats so that other teams can use the components without assistance.
• You will ensure operating system neutrality, making sure that tools and automations run on macOS, Linux, and Windows.
• You will keep an eye on costs and runtimes, identify inefficient patterns, and suggest viable alternatives.
• Required:
• You have several years of experience in software development across frontend and backend, and at least 10 years of experience in building frameworks, developer tooling, SDKs, or plugin systems for other development teams.
• You can design frameworks, platforms, and interfaces so that other teams can use them as a stable foundation, and you can convincingly justify technical and strategic decisions.
• You master schema and contract design with data models, interfaces, and extension points, for example, with JSON Schema and type systems, including versioning and backward compatibility.
• You have very good knowledge of TypeScript, especially type systems, generic types, interfaces, and plugin and adapter patterns, and you use Python for tools, checks, and evaluations, as well as portable shell scripts for automations.
• You have solid fundamentals in object-oriented design and common design patterns, including Builder, Factory, Strategy, Adapter, Repository, Pipeline, and Registry, and you are familiar with established architecture documentation formats.
• You have solid basic security knowledge regarding authentication, authorization, role-based permissions, secrets, trust boundaries, and typical threat scenarios like injection, Server Side Request Forgery, token leakage, and prompt injection.
• You have network and protocol knowledge of HTTP and HTTPS, retry and backoff strategies, rate limiting, and protocols with call semantics like JSON-RPC.
• You have experience with GitLab CI/CD and containers, build pipelines, anchor quality checks as blocking stages, and work securely in merge request-based processes.
• You work securely in Scrum teams, are familiar with backlog work, vertical slicing of tasks, Definition of Done, and have tool knowledge in Jira.
• You have been continuously using AI tools in software development for several months and use them reflectively in your daily work.
• You are familiar with the common terminology in the field of modern AI and agent systems and can classify their meaning in practical application.
• You can name the limitations, risks, and typical error patterns of AI technologies and derive well-founded deployment decisions from them.
• You have experience in designing and building agents with configuration layers such as agent definition, steering, skills, hooks, and guards.
• You have experience testing systems that do not deliver the same result every time and are familiar with suitable approaches for retries, comparison datasets, and handling unstable tests.
• You can pass on your knowledge in such a way that other teams can subsequently use the technical solutions independently.
• You are capable of acting without fundamental training in artificial intelligence topics.
• Desired:
• You have experience orchestrating multiple agents in production, including handovers, shared state, resumption after interruption, and handling loops.
• You have experience making the quality of results from agentic processes measurable and building and maintaining suitable evaluation metrics and comparison datasets.
• You are familiar with observability for AI systems, especially tracing across model calls, tool calls, and retrieval steps, as well as metrics for runtime, consumption, and costs.
• You have experience with sandboxing and isolation when executing code and tool calls and know the limits of such protection measures.
• You have experience with the Model Context Protocol or comparable protocols for connecting tools to AI assistants.
• You are familiar with Guardrail or Policy systems for technical security or behavioral control.
• You have ideally worked in a platform or enablement team whose internal customers are other teams.
• You are familiar with platforms for agentic services such as AWS Bedrock AgentCore, Microsoft Foundry Agents Service, or Microsoft Copilot Studio.
• You have experience in a scaled agile environment according to SAFe or comparable and are familiar with planning formats at a cross-program level.
• You ideally have experience in the logistics sector or a comparably complex corporate environment with high demands on security, data protection, and compliance.
• You have advantageous knowledge in Deno, in developing command-line tools for multiple operating systems, in embedded databases and full-text search, as well as English language skills.
• You convince with communication skills, teamwork, assertiveness, and an analytical mindset.
• Good German language skills, spoken and written (min. C1).
• A technically demanding senior role with a lot of creative freedom in building a reusable framework for an AI platform.
• An environment where you can make architecture-defining decisions and help shape the technical direction of a central platform.
• The opportunity to work on future-oriented topics related to agent systems, automation, platform engineering, and AI-powered development processes.
• An agile working environment with close collaboration between architecture, Scrum team, and user teams.
• An experienced and motivated team focused on quality, reusability, and sustainable technical solutions.
• Space to make your knowledge visible and empower other teams through documentation, demonstrations, and exchange formats.
• Personal and professional development opportunities in a modern technology environment.
• Service provision is 80% remote and 20% on-site. (Frankfurt (Main))

Project details

• Start: flexible
• Remote: hybrid
• Location: Frankfurt (Main), Deutschland, DE
• Languages: German