Agentic AI Engineer (m/f/d)

Description

The company runs CARA, its AI customer service assistant, which verifies customers, manages live reservations and answers policy questions across more than 10 markets in 7 languages. CARA is a production multi-agent system with write access to the booking platform, available on chat, app, WhatsApp and, soon, voice.

The company is a global mobility provider with revenue of €4.00 billion and about 9,000 employees. Its ONE platform combines the company rent, the company share, the company ride and the company+, with access to 350,000 vehicles, 4,000 cooperation partners and around 5 million drivers worldwide. Together with franchise partners it operates in more than 110 countries across 2,000 rental stations.

This role sits in the Data Science team and contributes to the customer service transformation. The work reaches millions of customers and is tied to measurable business outcomes.

Responsibilities

• You own your agents from start to finish: prompts, orchestration graphs, tools, evaluations, deployment and dashboards. You write the agent logic and review your own production traces.
• You turn business problems into agent features by defining tool calls and sub-agents with product, business and customer service stakeholders. You build for live transactions, including step-up authentication, guardrails, idempotent case handling and handover into Salesforce and Genesys.
• You make quality measurable by extending the evaluation harness with LLM-as-judge rubrics, tool-trajectory assertions and prompt A/B suites that must pass before each release.
• You optimise cost and latency alongside accuracy, using multi-provider model routing with fallbacks, prompt caching, and smaller models where they perform well enough.
• You work with software engineers, data scientists, MLOps and platform teams, weighing custom agent logic against shared platform components. You share agentic and LLM engineering knowledge and help other teams add Agentic AI to their products.
• Stack: Python with LangGraph and LangChain, containerised agents on AWS Bedrock AgentCore, MCP sub-agents, OpenAPI tool federation, and multi-provider fallback chains across Anthropic and OpenAI models behind Bedrock Guardrails. Next up is CARA for voice.

Requirements

• You are an engineer first, with solid hands-on software engineering experience, real production ownership, strong Python skills and good AWS cloud experience.
• You have built tool-calling agents with LangGraph, LangChain or similar frameworks and deployed them to real users, not only prototypes.
• You work hands-on with frontier and open models such as Anthropic Claude, OpenAI, AWS Bedrock and open-source models like Llama. You can explain when to use each, and you treat prompt engineering as an engineering discipline: versioned, reviewed, evaluated and A/B tested.
• You have applied LLMOps in practice, including tracing, evaluations, guardrails, token and latency budgets, and continuous integration and deployment.
• Bonus: MCP, real-time voice AI (WebRTC, SIP, streaming STT/TTS), event-streaming pipelines, TypeScript/React, or contact-centre platforms such as Genesys and Salesforce.
• You are highly motivated, eager to learn and practical about getting things done. You communicate fluently in English and value a diverse workforce.

Conditions

Two office days per week, with flexible working hours.

Equal opportunity: the company values diversity and welcomes all applicants.

Benefits

• Work in the Lisbon tech hub, at Factory Lisboa in Beato or ARTs Lisboa in Parque das Nações, alongside engineering, product and business teams.
• Regular internal and external speaker events for learning and knowledge sharing across teams and borders.
• Modern offices designed for both focused work and collaboration.
• 28 vacation days per year, an extra day off on your birthday, one volunteer day a year, and private health insurance that also covers your spouse and children at no cost.
• A yearly Coverflex allowance and a daily meal allowance.
• Exclusive rates on all the company services, with special discounts for family and friends.