Solution Architect

Select how often (in days) to receive an alert:

Solution Architect - Agentic Reporting 136735

Who Are We? Let me introduce us. You’re reading about Zurich Insurance Group , a global leader in insurance, but this role specifically pertains to our Technology Delivery Center (TDC). Established in Barcelona in 2006, the TDC operates as an integral part of Zurich Insurance Group, providing cutting-edge technological solutions and support to Zurich's local and global business units.

In essence, Zurich TDC is where technology meets the vast reach of a multinational company. We specialize in areas such as software development, data analytics, cybersecurity, and IT infrastructure. Our primary mission is to support Zurich's strategic goals and enhance operational efficiency through innovative technology solutions.

If you're considering a role with us, know that you'll be joining a hub of technological excellence dedicated to driving Zurich's success on a global scale.

What Can You Expect? The Solution Architect is accountable for the end-to-end technical design of ZAAi (Zurich’s Actuarial Agentic Application) — a Group-wide agentic AI platform for actuarial and financial reporting, built on a multi-agent orchestration framework that connects LLMs to deterministic calculation engines via structured tool-use (MCP).

This role combines deep domain understanding of actuarial and finance applications, agentic AI solution design, and enterprise delivery to ensure the platform connects the right data to the right agents in a governed, auditable, and scalable way. The starting scope is non-life actuarial reporting, with planned expansion into Life, Reserving, and Group Finance.

And you might be thinking: This is great, what will my day-to-day be like, at TDC? I’m glad you asked.

As a Solution Architect — Agentic Reporting you will have the opportunity to:

Develop deep understanding of Zurich’s actuarial and finance applications — their data models, calculation outputs, and how results flow from Business Unit through Region to Group — and use that understanding to design what the agentic reporting solution should do and with which data.

Design the data integration layer: how structured outputs from source systems feed into the agentic framework in a governed, traceable, and reproducible way.

Define the solution architecture across the full stack — data connectors, MCP tool-use, LLM orchestration, deterministic compute, domain specific expertise into LMM’s skill, memory persistency via DB — with a clear view of the outcome each component serves.

Agentic AI Platform Design

Define how LLM orchestration, tool-use (MCP servers), and deterministic compute layers connect and scale — balancing AI flexibility with auditability and cost control.

Apply the simplest architecture that meets the need — knowing when a single well-designed agent beats a complex multi-agent system.

Own non-functional requirements specific to agentic workloads: latency, token cost governance, model fallback, reproducibility, and observability.

Translate business workflows into agentic solution designs: which steps can be automated, which data sources to connect, what the agents need to know, and where human review is required.

Own the evaluation framework for agent outputs: eval sets, tracing (e.g. Langfuse / OpenTelemetry), and cost/latency/hallucination monitoring.

Act as the bridge between actuarial/finance domain experts and the technical delivery team (Reconverz, IT infra, data engineers).

Act as the technical interface between Zurich IT, the Reconverz application vendor, and internal data and dashboard teams.

Clearly articulate trade-offs, impacts, and architectural recommendations to senior stakeholders.

Governance & Expansion

Ensure the platform meets Zurich IT governance, CCoE, and architecture standards.

Own AI security and responsible-AI guardrails: prompt-injection defence, safe tool execution, PII/secrets handling, and human-in-the-loop controls for high-risk agent actions.

Lead the expansion of the solution architecture beyond non-life actuarial into Group Finance reporting — identifying reusable patterns and where domain-specific customisation is needed.

Contribute to the control and evaluation framework: defining how agent outputs are validated against source data and business expectations.

Work

alongside experts in various technological domains.

Enhance

your skills with real-world experience.

Contribute

to a team that values your input and creativity.

What Makes You a great Fit? While we’re not seeking perfection or expecting you to save the world, we believe that if you meet some of these requirements, you’ll be a fantastic fit for our team. We are waiting for