Forward Deployment Architect

Job Purpose

Xebia is building in-house AI expertise to deliver AI in aviation. We build; we do not buy off the shelf. As a Forward Deployed AI Engineer, Technical Lead you set the technical direction for a squad embedded in the business: you sit with stakeholders to understand the problem and the why , then design, build, deploy and run the agentic AI systems that solve it, end to end. This is a hands-on, build-first role with single-threaded ownership of a real aviation outcome, working alongside a Business Product Owner and an AI Value Architect, on a shared platform (paved road, MCP fabric, standards) that lets your squad self-serve against Customer’s systems.

Accountabilities & Responsibilities

Understand before you build. Start every problem with the business need and the why, working directly with stakeholders, then take the solution from discovery to production.

• Design, build, deploy and continuously improve enterprise-grade agentic AI applications for real aviation scenarios, using agentic coding as your default way of working.
• Build agents that reason across steps, call tools and APIs, manage context, handle exceptions and support human-in-the-loop, reliably and at enterprise scale.
• Design and implement RAG pipelines over enterprise knowledge: ingestion, chunking, embeddings, vector search, retrieval tuning, grounding and source traceability.
• Build MCP-based integrations and connect agents to backend systems via REST/OpenAPI, webhooks and event-driven patterns with secure authentication, and expose your own work as clean, reusable, self-serviceable interfaces.
• Apply structured LLM patterns end to end: tool calling, schema-validated outputs, retries, fallbacks and guardrails.
• Own quality from day one: testing, evaluation, observability, logging, versioning and feedback loops for reliability, accuracy, latency, security and cost.
• Apply security, privacy, access control, auditability, responsible-AI and governance across every deployment.
• Take single-threaded ownership of a domain outcome (one owner, one result) and help establish reusable patterns that grow Customer’s internal AI capability rather than renting it.
• Coordinate with your Business Product Owner, AI Value Architect and other squads; speak up when AI is not the right tool.
• Set the technical direction and standards for the squad’s agentic AI work, and make the key architecture and build-vs-buy calls.
• Design multi-agent and agent-to-agent systems and evaluation frameworks that keep them reliable, and lead delivery with external AI platforms and vendors while building Customer internal capability.
• Grow the engineers around you: mentor, review, and raise the bar on quality, security and cost across the squad.

Education & Experience

We look for a technical lead who sets the engineering direction for agentic AI while still building, and combines that with a business-first mindset:

• Curiosity above all: you dig into problems, question assumptions and want to understand how the airline actually works.

• A business-first, human-centric mindset: aviation is made for humans, by humans, and AI supports people, it does not replace them. Fluent English, comfortable in a culturally diverse, international team.

• 8+ years building production-grade software, including 4+ years with GenAI, LLMs and applied ML and at least 1 year of hands-on agentic AI as an early adopter, with a track record of setting technical direction and shipping agentic systems at scale.

• Hands-on experience or strong working knowledge of MCP (Model Context Protocol) for connecting agents to tools, systems, APIs and data.

• Strong Python, with strong knowledge of at least one of TypeScript / JavaScript, Java, or C#, and modern engineering practice: async programming, FastAPI, Pydantic, Git and CI/CD, testing, error handling and logging.

• Practical experience with at least one agent framework or enterprise AI platform (e.g. LangGraph, Semantic Kernel, CrewAI, AutoGen, OpenAI Agents SDK, Microsoft Foundry, Amazon Bedrock AgentCore, Google Vertex/Gemini) and with a vector database or search platform (e.g. Azure AI Search, pgvector, Pinecone, Weaviate, OpenSearch).

• Experience integrating enterprise systems (APIs, managed identities, webhooks, queues, middleware) and deploying on cloud with containers, monitoring and observability; sound judgement on the trade-offs of latency, quality, cost and reliability, and on security, privacy, responsible AI and governance.

• Strong assets: aviation or airline domain knowledge; a background in classical machine learning and data science; and classical full-stack development (interfaces, frontends, APIs, backend engineering).

• Preferred for this level:

• Experience building AI agents for complex enterprise or operations-heavy workflows (logistics, supply chain, aviation, cargo, customer operations or contact centre).

• Experience with voice AI, email automation, CRM integrations, workflow automation or multilingual agents.

• Experience designing golden test sets, simulation-based testing, regression testing and agent evaluation frameworks.

• Experience designing multi-agent systems and agent-to-agent communication patterns, agent registries or tool-orchestration standards.

• Experience leading delivery with external AI platforms, startups or vendors while building internal engineering capability.

• Master’s degree in computer science, Software Engineering, Data Science, AI/ML or a related technical field, or equivalent practical experience; relevant cloud-AI, GenAI, agentic-AI or MLOps certifications are an advantage.

• Growth path: the natural next step is AI Value Architect, owning a cluster’s value journey while still building alongside the team.