AI engineer in regulatory technology

Reviva | Deorganising Crime ™; Cascais; On-site; EUR 101,000 - 120,000; Full time

Job summary

Reviva | Deorganising Crime ™ ищет Senior Agentic AI Engineer для разработки новой AI-возможности в рамках существующей SaaS‑платформы. Ваша задача — спроектировать архитектуру для клиентского AI‑помощника, который взаимодействует с сервисами продукта, данными и рабочими процессами.

Нужен практический опыт вывода агентов в продакшн и сильный бэкграунд в Python и/или .NET, API, распределённых системах и облачной инфраструктуре.

Qualifications

• Практический опыт разработки производственных AI-агентов.

Responsibilities

• Проектирование многошаговых рабочих процессов агентов.
• Вызов инструментов и функций через существующие API/серверные сервисы.
• Управление состоянием и контекстом.
• Обработка повторов, откатов и восстановления после сбоев.
• Контроль за человеческим участием в процессе.
• Извлечение данных как структурированных, так и неструктурированных.
• Оценка и регрессионное тестирование.
• Трассировка, наблюдаемость и отладка в проде.
• Версионирование подсказок и моделей.
• Гарантия соответствия, разрешения и аудитируемость.
• Управление стоимостью токенов и задержками.
• Интеграция агентного ИИ в существующую SaaS-платформу.

Skills

Production-grade AI agents

Agentic architecture design

Python

.NET / C#

APIs

Distributed systems

Cloud infrastructure

Databases

Tools

LangGraph

Semantic Kernel

Pydantic AI

Azure AI Foundry

Job description

Описание:

A growing B2B SaaS business in the regulatory technology space is hiring a Senior Agentic AI Engineer to help design and build a new AI capability across its core product. The business already has an established software platform and experienced engineering team. This contract will focus on shaping the architecture for a customer-facing AI assistant that can interact with existing product services, structured data, documents and workflow logic. The longer-term product direction includes helping customers create and manage complex regulatory workflows through conversation, interpret changing requirements and use existing platform capabilities more intelligently.

Задачи:

• Designing multi-step agent workflows


• Tool and function calling across existing APIs and backend services


• State and context management


• Retry, fallback and failure-recovery behaviour


• Human-in-the-loop controls


• Retrieval across structured and unstructured data


• Evaluation and regression testing


• Tracing, observability and production debugging


• Prompt and model versioning


• Guardrails, permissions and auditability


• Token-cost and latency management


• Integrating agentic AI into an existing production SaaS platform

Требования:

• You’ll need strong hands‑on experience building AI agents that have reached production. You should be comfortable discussing: how your agent architecture was structured, how tools and APIs were exposed, how state was managed, how failures and retries were handled, how outputs were evaluated, how bad runs were traced and debugged, how cost and latency were controlled, what you learned once the system was being used in production


• A strong software engineering background is essential. Experience with Python and/or .NET, APIs, distributed systems, cloud infrastructure and databases will all be relevant.


• Right to work in Portugal required

Будет плюсом:

• Experience with LangGraph, Semantic Kernel, Pydantic AI, Azure AI Foundry or similar frameworks would be useful;


• Experience in insurance, financial services, compliance, GRC or another regulated environment would also be valuable

Условия:

€550–€650/day Initial 3–6 month engagement High level of ownership across architecture and implementation Direct collaboration with the core engineering team No sponsorship available