Amazon's Finance Technology group is seeking a Software Development Engineer specializing in FinTech to join our team. You will be instrumental in creating advanced machine learning and agentic-AI systems designed to analyze billions of dollars in daily transactions. Your work will directly contribute to significant cost savings by defending Amazon against theft, fraud, abuse, and waste across various operational domains.

We are at the forefront of building the next generation of fraud detection platforms, leveraging generative and agentic AI. This involves developing multi-agent systems capable of identifying complex fraud patterns directly from raw data and collaborating with domain experts to translate these findings into production-ready detection logic. This innovative approach aims to dramatically accelerate the deployment of new detection capabilities, moving from sprints to days.

As a key member of this team, you will design and implement agent tools, ML pipelines, services, and the underlying data infrastructure. Your role will involve working with cutting-edge technologies including generative AI, large language models, and agent frameworks, in addition to established techniques like anomaly detection, classification, and graph modeling. You'll have the opportunity to partner closely with applied scientists, fellow engineers, and finance operations teams, witnessing your innovative ideas come to fruition rapidly.

Engage in the complete software development lifecycle, from conceptualization and customer collaboration to the deployment of scalable, production-ready systems.

Design and construct robust ML pipelines capable of processing vast datasets and scoring billions in transaction value.

Develop sophisticated agentic AI applications, including LLM-powered agents that uncover risk patterns and generate detection logic through well-defined interfaces with human oversight.

Build critical agent support systems such as vector stores, knowledge graphs, entity-risk layers for collusion detection, and various memory components.

Establish metrics and methodologies for evaluating agent performance in fraud detection and translate analyst investigation patterns into repeatable, templated workflows.

Create the platform and tooling necessary for safe and rapid deployment of detection logic by both agents and non-engineers, encompassing governed data access, model deployment, and self-service rule authoring.

Possess a minimum of 3 years of professional software development experience, excluding internships.

Accumulate at least 2 years of experience in designing or architecting new and existing systems, with a strong understanding of design patterns, reliability, and scaling.

Demonstrate proficiency in programming with at least one software programming language.

Preferred qualifications include 3+ years of experience across the full software development lifecycle, encompassing coding standards, code reviews, source control, build processes, testing, and operations. A Bachelor's degree in Computer Science or a related field is beneficial. Experience in machine learning, data mining, information retrieval, statistics, or natural language processing is highly valued. Familiarity with massively parallel data processing technologies such as Redshift, Spark, or Hadoop-based solutions is also advantageous.