AI Product Engineer – FLOW HUB

What's the project's context?

Development of a central enterprise platform that acts as the control tower for AI agents and automation solutions across the organization.

The platform covers the full lifecycle — from use case intake via an ideation portal and guided building in a development studio, to compliant release through a governed deployment pipeline (ensuring adherence to legal and corporate standards like the EU AI Act).

It also includes agent monitoring with usage, token, and cost transparency in a central catalog.

The platform integrates a heterogeneous landscape of technologies (such as Microsoft, SAP, ServiceNow, and other third-party providers) behind a single, platform-independent discovery and orchestration layer.

Your role

We are seeking an AI Product Engineer to support the development of a central enterprise platform that serves as the control tower for AI agents and automation solutions, enabling the end-to-end lifecycle from ideation and development to compliant deployment, monitoring, and orchestration across a heterogeneous technology landscape.

Most important/mandatory skills:

• very strong frontend skills with React and TypeScript (Next.js), incl. component architecture, state management and performance
• strong backend skills with Java and/or Kotlin and Spring Boot — REST APIs, persistence, authorization
• hands-on experience with LLM-based applications: prompt architecture, RAG, tool/function calling, agent patterns
• working knowledge of Python for data preparation, evaluation and prototyping
• experience with agile software development in large enterprise projects
• ability to work directly with business stakeholders — requirement elicitation, demos, technical trade-off discussions
• fluent English & German (written and spoken)

Optional skills:

• Spring AI (chat clients, advisors, tool calling, vector store abstraction)
• MCP and/or A2A protocols, agent orchestration frameworks
• Azure and/or AWS, Kubernetes, Docker, CI/CD (Maven/Gradle, GitHub Actions)
• vector databases (pgvector), embedding and chunking strategies, reranking
• LLM evaluation and observability — grounding checks, regression suites, tracing, token monitoring
• Spring WebFlux / reactive streaming endpoints for LLM responses
• design system affinity (Figma, design tokens, accessibility)
• Bosch enterprise landscape: OneIDM, LeanIX, ServiceNow, Power BI, MS Graph

Main task/activity

Development of user-facing AI product features on the FLOW HUB platform — from prototype against real business data through to production — with focus on the React/Next.js frontend and the Spring Boot backend services for agent discovery, orchestration and governance.

Other tasks and activities

• Implementing and extending the FLOW HUB frontend (Next.js/React, TypeScript): agent catalog and search, guided build journey, governance and monitoring dashboards
• Designing and implementing UI patterns for non-deterministic AI interactions — streaming responses, agent run traces, tool-call visualisation, source attribution, human-in-the-loop approvals
• Implementing backend services in Kotlin/Java with Spring Boot, including the API manager services for Agent Search, Agent Sync and Tool Recommender
• Integrating LLM capabilities via Spring AI: prompt architecture, RAG pipelines against Bosch knowledge sources, tool/function calling, multi-step agent orchestration
• Connecting third-party agent platforms and enterprise systems (Copilot Studio, DIA Brain, Cognigy, n8n, ServiceNow, LeanIX, Power BI) and the identity layer (OneIDM, Compas, MS Graph)
• Rapid prototyping of new AI product features together with the requesting departments against their real data and workflows, and hardening validated prototypes into production features
• Building evaluation suites and observability for LLM features — grounding, regression and hallucination checks, latency, token consumption and cost per agent
• Collecting and refining requirements with product owners, process owners and GSOs; deriving user stories and contributing to the product backlog
• Implementing and documenting compliance-relevant features: EU AI Act readiness, CD0302 release process, role and permission model
• Supporting containerized deployment, CI/CD pipeline and operations, including defect analysis and iteration after go-live