Build AI capabilities that make insurance documents and requests easier to work with. Your focus will be useful outputs, clear source context, measurable quality, and reliable handoffs to human reviewers.
What you’ll do
- Develop document extraction, classification, summarization, and question-answering workflows.
- Build retrieval workflows that connect responses with relevant, approved source material.
- Design structured model outputs and validate them before downstream actions.
- Create representative evaluation datasets and measure quality, failure cases, latency, and cost.
- Implement uncertainty handling, prompt-injection defenses, access-aware retrieval, and human review paths.
- Partner with software engineers and insurance specialists to improve models and workflows using reviewed feedback.
What you’ll bring
- Strong Python skills and experience integrating model APIs or deploying machine-learning applications.
- Practical understanding of large language models, embeddings, retrieval, and structured outputs.
- Ability to build repeatable evaluations and distinguish a convincing response from a supported result.
- Experience processing documents or text data and handling incomplete or noisy inputs.
- Good software engineering habits, including version control, tests, documentation, and careful data handling.
Helpful experience
- Experience with retrieval-augmented generation, vector search, OCR, or document parsing.
- Familiarity with model monitoring, tool use, and controlled workflow execution.
- Experience with insurance, financial-services, or other document-heavy business processes.
What good work looks like
An evaluated workflow that produces useful, source-supported results, makes uncertainty visible, and routes the right cases for human review.
Include with your application
Share an AI project and explain how you evaluated its quality. Include a failure case, how you detected it, and what you changed. Use public or non-confidential examples.
Role location, work arrangement, employment terms, and compensation will be confirmed during the hiring conversation.
