Applied Scientist, FinTelligence
At Amazon's FinTech organization, we are building AI systems that process hundreds of millions of financial transactions, turn complex documents into actionable intelligence, and power autonomous agents that learn from every customer interaction. We are looking for an Applied Scientist to lead the development of generative AI applications that change how finance teams work, tackling problems at the intersection of large language models, multi-agent systems, and real-world financial operations.
Key job responsibilities
- Building AI systems that finance teams trust enough to rely on without manual review, where precision isn't a nice-to-have, it's a compliance requirement.
- Designing agents that learn from user corrections and get measurably better with every interaction, not just at the next model release.
- Solving inference at massive scale using tiered model architectures, intelligent routing, and small language models that deliver production-grade accuracy at a fraction of frontier model cost.
- Developing evaluation frameworks that catch quality regressions before customers do and gate every model change before it ships.
Who Thrives Here
- You're someone who cares as much about shipping as about research.
- You've built models that run in production, not just in notebooks.
- You're comfortable working across the full stack, from model architecture to deployment to measuring whether the customer's workflow actually changed.
- You operate well in cross-functional settings where science, engineering, and business teams inform each other continuously.
- You'd rather solve a hard real-world problem than optimize a benchmark.
What Makes This Different
- Your work ships to production and directly changes how thousands of finance professionals operate daily.
- The problems are genuinely hard: financial data is messy, regulated, high-stakes, and operates at a scale where naive LLM approaches break down.
- You'll work across multiple domains, from contract intelligence to cash application to financial data investigation, not a single narrow use case.
Key job responsibilities
- Building AI systems that finance teams trust enough to rely on without manual review, where precision isn't a nice-to-have, it's a compliance requirement.
- Designing agents that learn from user corrections and get measurably better with every interaction, not just at the next model release.
- Solving inference at massive scale using tiered model architectures, intelligent routing, and small language models that deliver production-grade accuracy at a fraction of frontier model cost.
- Developing evaluation frameworks that catch quality regressions before customers do and gate every model change before it ships.
Who Thrives Here
- You're someone who cares as much about shipping as about research.
- You've built models that run in production, not just in notebooks.
- You're comfortable working across the full stack, from model architecture to deployment to measuring whether the customer's workflow actually changed.
- You operate well in cross-functional settings where science, engineering, and business teams inform each other continuously.
- You'd rather solve a hard real-world problem than optimize a benchmark.
What Makes This Different
- Your work ships to production and directly changes how thousands of finance professionals operate daily.
- The problems are genuinely hard: financial data is messy, regulated, high-stakes, and operates at a scale where naive LLM approaches break down.
- You'll work across multiple domains, from contract intelligence to cash application to financial data investigation, not a single narrow use case.