Software Development Engineer, FinTech TFAW

Do you want to build machine learning and agentic-AI systems that analyze billions of dollars in transactions every day and measure your impact in millions of dollars saved

In Amazons Finance Technology group we build the detection systems (machine learning and increasingly agentic AI) that defend Amazon against theft fraud abuse and waste across the supply chain retail catalog and corporate procurement. We turn terabytes of transaction vendor and operational data into real-time risk decisions: scoring payments as they happen surfacing anomalous behavior and uncovering collusion networks that span thousands of entities.

Were now building the next generation of that platform around generative and agentic AI: multi-agent systems that discover fraud patterns directly from the raw data flowing through Amazons businesses working alongside domain experts to turn what they find into production detection logic. The result is a platform where new detection coverage comes not only from engineers but from agents and the experts who know the fraud best shipping in days instead of sprints.

Youll build the agent tools ML pipelines services and data infrastructure behind all of this working with generative AI large language models and agent frameworks alongside the anomaly-detection classification and graph-modeling techniques they build on. Youll partner directly with applied scientists fellow engineers and finance operations teams and youll see your ideas reach production in weeks not years.

If you want to apply advanced ML and agentic AI at Amazon scale on systems with a direct measurable financial impact wed love to talk.

Key job responsibilities
You will participate in the full software development lifecycle from collaborating with customers and scientists on design to building scalable extensible systems that run in production. You will:

Design and build ML pipelines that process terabytes of data and score billions of dollars in transactions.
Build agentic AI applications: LLM-powered agents that discover risk patterns and generate detection logic operating through well-defined tool interfaces with human-in-the-loop review.
Build the knowledge and memory systems agents depend on: vector stores knowledge graphs the entity-risk layer for cross-entity collusion detection and in-session and shared memory.
Develop how we evaluate agent effectiveness in fraud detection and turn analysts ad-hoc investigation patterns into repeatable templated workflows.
Build the platform and tooling (governed data access model deployment and self-service rule authoring) that lets agents and non-engineers ship detection logic safely and fast.
Take models and agents from prototype to reliable real-time production systems and work directly with internal customers to turn feedback into shipped improvements.

A day in the life
Youll partner with applied scientists software engineers and finance operations teams across Amazon. Youll harden an agent pipeline for production extend the entity graph that powers a new detector or design a tool interface that lets an agent query data safely then review it with the customer wholl use it. Because our team and our customers are comfortable trying new ideas youll get your work in front of real data quickly and iterate on what you learn.

About the team
We own the machine learning and agentic-AI systems that prevent recover and avoid internal and external theft fraud abuse and waste across Amazon Finance Operations. Our platform makes real-time risk decisions on transactions from many of Amazons largest businesses. Were actively building toward entity-level risk profiles updated in real time and agentic systems that compress fraud discovery and investigation from weeks to days with engineers building the leverage layer that multiplies what every analyst and agent can do.

- 3 years of non-internship professional software development experience
- 2 years of non-internship design or architecture (design patterns reliability and scaling) of new and existing systems experience
- Experience programming with at least one software programming language

- 3 years of full software development life cycle including coding standards code reviews source control management build processes testing and operations experience
- Bachelors degree in computer science or equivalent
- Experience in machine learning data mining information retrieval statistics or natural language processing
- Experience in processing data with a massively parallel technology (such as Redshift Teradata Netezza Spark or Hadoop based big data solution)

Our inclusive culture empowers Amazonians to deliver the best results for our customers. If you have a disability and need a workplace accommodation or adjustment during the application and hiring process including support for the interview or onboarding process please visit for more information. If the country/region youre applying in isnt listed please contact your Recruiting Partner.