SlipstreamJobs tracks this role from the company's public career site. Apply directly on the employer's site.
mea is building an intelligent operating layer for the insurance industry, combining a domain-specific language model with an insurance knowledge graph to automate workflows end-to-end. Founded in 2021 and bootstrapped to profitability for 4 consecutive years, the company raised $50 million in February 2026 and now serves clients across 21 countries. The team is fully remote and global, spanning over 100 insurance veterans and technologists across the UK, US, India, continental Europe, and the Middle East.
You will join the AI team as a Prompt Engineer, designing, testing, and continuously improving prompts and AI workflows to extract, structure, and validate data from complex, unstructured documents. This role sits at the intersection of AI engineering, domain expertise, and product development.
Key Responsibilities:
- Design, develop, and optimize prompts for LLM-based workflows including extraction, question-answering, and validation tasks
- Perform field-level extraction logic across complex documents (PDFs, emails, Excel files, etc.)
- Analyze model outputs to identify failure patterns, edge cases, and inconsistencies
- Continuously improve prompt accuracy through iterative testing and refinement cycles
- Collaborate with domain experts to understand business context and data nuances
- Support confidence calibration and validation logic
- Contribute to building reusable prompt templates and libraries
- Work with structured outputs (JSON, tables, schemas)
- Assist in integrating feedback loops including human-in-the-loop corrections and RLAIF
- Document prompt strategies, assumptions, and improvements
Requirements:
- 2–5 years of experience required
- Strong understanding of LLMs and prompt engineering concepts
- Ability to work with JSON, structured outputs, and schema-driven extraction
- Strong analytical skills to debug model outputs
- Familiarity with prompt chaining, context management, few-shot and zero-shot techniques
- Ability to break down ambiguous requirements into structured logic
- Strong attention to detail (critical for extraction accuracy)
- Pattern recognition across different document formats
- Ability to handle edge cases and exceptions
- Clear communication of issues, findings, and improvements
- Ability to work effectively with AI engineers, domain experts, and product teams
- Comfortable documenting learnings and prompt strategies
- Must be authorized to work in India
Nice-to-Have Skills:
- Experience with LangChain, LlamaIndex, or similar frameworks
- Basic Python knowledge for testing and scripting
- Exposure to document processing (OCR, PDFs, etc.)
- Understanding of insurance domain (underwriting, claims, SOV, loss runs)
- Familiarity with evaluation metrics (accuracy, precision, recall, confidence)
- Experience with annotation and labeling tools
