machine learning engineer for payment products

Adyen provides payments, data, and financial products in a single solution for customers such as Meta, Uber, H&M, and Microsoft. Its Payment Solutions team processes billions of transactions and helps businesses create seamless payment experiences through a global financial technology platform.

*Instagram и Facebook принадлежат компании Meta Platforms Inc., деятельность которой признана экстремистской и запрещена на территории РФ

• Discover critical challenges across product and engineering teams and rapidly build AI prototypes to demonstrate value;
• Own the end-to-end development of bespoke AI tools for merchant experience, pricing models, and internal workflows;
• Define and lead evaluation strategies for agentic systems and LLMs;
• Design internal benchmarks covering domain complexity, edge cases, capabilities, and failure modes;
• Build reusable evaluation infrastructure embedded in the development process;
• Provide technical expertise on agentic frameworks, retrieval and search strategies, and agent tool-use approaches across partner teams;
• Identify connections across AI initiatives and help teams avoid duplicated work or incorrect approaches;
• Set engineering standards for the team and company;
• Mentor through problem decomposition, research methodology, and code review;
• Promote reproducibility, documentation, and rigorous evaluation practices across the AI organization.

• 7+ Years of hands-on experience in applied AI/ML research or engineering;
• A proven track record of shipping AI systems, including agentic or LLM-powered systems, in production;
• Deep expertise in language models and Generative AI;
• Hands-on experience with architecture, post-training, inference optimization, context engineering, and failure modes at scale;
• Experience designing and operating agentic systems at scale, including multi-agent orchestration, tool use, memory and context management, state handling for long-running workflows, and human-in-the-loop design;
• Experience designing evaluation frameworks or internal benchmarks beyond standard metrics;
• Understanding of LLM-as-judge failure modes and meaningful system evaluation;
• Strong foundation in supervised learning, ensemble methods, optimization, probabilistic modeling, and statistics;
• Ability to write clean, well-structured, production-ready Python code;
• Hands-on experience with at least one production-grade agentic framework;
• Nice to have: Familiarity with financial data, payments, fraud detection, or risk systems, publications, conference presentations, open-source contributions, observability and evaluation tooling, MLOps, model deployment pipelines in large-scale environments.

• The role is based out of the Amsterdam office;
• The company is office-first and values in-person collaboration;
• Remote-only work is not offered.