Building AI-powered products across the health insurance domains - spanning claims assessment, underwriting, policy management, customer servicing, fraud detection, and more. Primary AI stack is Google Gemini (Vertex AI), and project will involve actively building AI agents that automate and augment complex insurance workflows end-to-end.
This is a technology-first team. You will work on real AI integration challenges - prompt engineering, agentic workflows, multi-step reasoning, tool-use, structured LLM outputs - in a production health insurance environment. Domain knowledge in health insurance is a bonus
What you will do:
- Build AI-powered features across the health insurance platform using Google Gemini on Vertex AI
- Design and optimize prompts for complex multi-step reasoning tasks: analysis, classification, extraction, and decision support
- Work with Gemini's structured output capabilities (JSON mode, function calling, grounding) to produce reliable, deterministic results
- Build and iterate on AI agents - multi-turn workflows where the model uses tools, calls APIs, and makes sequential decisions autonomously
- Manage context windows efficiently by building context builders that supply the right data without bloating the prompt
- Evaluate model outputs, measure accuracy and consistency, and improve reliability through better prompting, guardrails, and validation
- Stay current with Gemini model releases (Flash, Pro, Ultra) and apply the right model per use case
What we are looking for:
- Hands-on experience with Google Vertex AI / Gemini API - personal projects and hackathons count
- Understanding of prompt engineering: few-shot, chain-of-thought, structured output, tool/function calling
- Ability to design agentic workflows - knowing when to chain LLM calls, when to use tools, and how to handle failures
- Familiarity with Node.js or Python for AI service backends
- Strong reasoning skills - ability to read an LLM output and diagnose exactly why it went wrong
- Health insurance domain knowledge: add-on, not mandatory
Common Expectations
- Tech-first mindset - you are here to build AI systems, not to become an insurance expert
- Genuine curiosity about AI agents - you find it interesting to think about how LLMs reason, where they fail, and how to build reliable systems around them
- Ownership - you will own your feature end-to-end, from design to production
- Collaborative - small team, fast-moving, everyone's work touches everyone else's
- Communication - ability to explain AI system behaviour and limitations to non-technical stakeholders
Tech Stack
AI/LLM: Google Gemini (Vertex AI) - primary stack; OpenAI GPT-4 / Claude used for benchmarking
AI Patterns: Prompt engineering, function calling, AI agents, tool use, structured output, grounding
Backend: Python/Node/JAVA
Database: PostgreSQL, Bigquery
API: REST, GraphQL
Cloud: GCP (Cloud Run, Vertex AI, GCS)
CI/CD: Git, Docker
