Agentic AI Engineer

In this role you will enable Sunrise to design, deploy and continuously improve production‑grade agentic and conversational AI solutions across customer and employee workflows. You will provide shared frameworks, standards and quality gates to ensure AI behaviour is predictable, grounded in governed enterprise knowledge and data, safe, multilingual and operationally ready. You will partner with ADAO Domain Pods and engineering teams to accelerate delivery through reusable prompt/agent patterns, evaluation harnesses and disciplined release management. YOUR CHALLENGE: Agentic Systems Architecture: Design and evolve enterprise agent frameworks including orchestration, tool/function calling, workflow integration, retrieval/grounding and state management. Develop reusable agent infrastructure (“Agent Factory”) consisting of modular agent templates, tool adapters, prompt modules, evaluation pipelines, and reference implementations to enable scalable and consistent deployment. Engineer structured single-agent and multi-agent systems with clearly defined reasoning boundaries, escalation paths and autonomy levels. Define architectural patterns for distributed agent workflows across domains and business Pods. Reliable LLM Engineering: Implement reliability patterns for LLM-based systems including fallback strategies, structured output validation, deterministic routing and failure containment mechanisms. Optimize latency, token efficiency and cost-per-interaction while maintaining quality thresholds. Apply context management and grounding techniques to ensure predictable and auditable behavior. Prompt Engineering & Conversation Logic: Engineer high-quality prompts and system instructions for safe, grounded and structured agent behavior. Define and maintain prompt libraries, versioning standards and regression testing practices. Design multilingual (DE/FR/IT/EN) conversational flows aligned with enterprise and regulatory standards. Evaluation & Quality Control: Design and maintain evaluation harnesses including regression suites, golden conversations and structured scoring frameworks Define measurable acceptance thresholds and release readiness criteria. Monitor production behavior including grounding quality, hallucination risk, escalation patterns and performance metrics. Production Integration & Observability: Integrate agentic systems into enterprise channels and backend services with robust authorization, state handling and error management Implement observability standards including traces, logs, telemetry and structured evaluation reporting Support go-live, hypercare and post-incident improvement cycles. Safety, Risk & Governance: Implement bounded autonomy and human-in-the-loop control patterns. Enforce guardrails, policy compliance and PII protection mechanisms. Collaborate with Security, Privacy and Governance teams to ensure enterprise-grade compliance. Examples of what you’ll help build: AI Agents to Enhance Customer Operations: Design and implement intelligent agents that analyze customer requests, identify root causes, and suggest effective resolution paths, helping improve service quality, consistency and operational efficiency. Expansion of an Enterprise Multi-Agent Framework: Develop and integrate new specialized agents into a scalable multi-agent architecture to automate workflows, support internal teams, and continuously improve operational processes. Deployment of AI Copilots for Internal Teams: Work closely with business stakeholders to introduce task-focused AI copilots that assist with activities such as case analysis, information retrieval, and process coordination, reducing manual effort and accelerating turnaround times.

At ADAO (AI, Data and Agentic Enablement Office), we bring people together to unlock the future of Sunrise through data, analytics, AI, and agentic technologies. We think big, move with purpose, and turn ideas into real impact for our customers and teams. ADAO is where innovation meets trust: we build strong data foundations, scale safe and reliable AI capabilities, and empower business units through a hub‑and‑spoke operating model to deliver meaningful results across the company. Within this model, you will work as part of the Agentic Enablement and Prompt Engineering & Conversation Design team, where the central hub provides standards, platforms, architecture, and best practices, while cross‑functional, outcome‑driven Pods are embedded in and led by the business. You will act as an engineering expert, partnering closely with the business to deliver on their priorities and outcomes.

• Bachelor’s or Master’s degree in Computer Science, Engineering, AI or equivalent practical experience

• Strong hands-on experience building and deploying LLM-based applications or agentic systems in production

• Deep understanding of: Tool/function calling and structured outputs, Retrieval-Augmented Generation (RAG), Multi-step workflows and orchestration logic, Context management and reliability techniques for LLM systems

• Experience implementing evaluation frameworks and regression testing for AI systems

• Experience integrating AI services into backend systems and cloud-native environments

• Strong software engineering skills in Python, API design and containerized deployments

• Familiarity with observability, monitoring and production operations for distributed systems

• Ability to reason about system boundaries, autonomy levels and failure modes

• Experience working in cross-functional product teams in federated or hub-and-spoke delivery environments

• Fluent in English; German is a strong advantage