Senior Software Engineer

Senior Software Engineer, Agentic Core

About the Role

We're seeking a Senior Software Engineer to own the architecture and development of our Agentic Core—the foundational AI orchestration layer within Eicore. You will bridge our existing robust microservices with advanced LLM capabilities, building autonomous agent workflows that can handle complex insurance operations like claims triage, underwriting assistance, and policy querying.

Technology Stack

• Core: Bun, TypeScript, React (Vite, TailwindCSS, Shadcn/UI)
• Data: Prisma ORM, PostgreSQL (including pgvector for AI)
• AI/Agentic: Claude (Anthropic), OpenAI, Vercel AI SDK (or LangChain.js), Zod for structured LLM outputs
• Infra: Docker, GitHub Actions, DigitalOcean

Responsibilities

• Architect the Agentic Core: Design and build the TypeScript-based orchestration engine that manages agent planning, state, tool execution, and self-correction loops within the Bun monorepo.
• Implement Tool-Use/Function Calling: Design secure, standardized interfaces that allow AI agents to safely interact with EICore microservices (e.g., querying a policy via Prisma, initiating a claims workflow) using OpenAPI/Zod schemas.
• Build Advanced RAG Pipelines: Develop retrieval-augmented generation workflows using pgvector and embeddings, allowing agents to securely query complex policy documents, underwriting guidelines, and claims history.
• AI Observability & Evaluation: Implement deep tracing, logging, and monitoring for agent workflows. Build automated evaluation frameworks to measure agent accuracy, hallucination rates, and cost-efficiency.
• System Architecture: Define shared schema contracts (Zod + OpenAPI) that ensure type safety across LLM boundaries, service layers, and UI components.
• Infrastructure & Deployment: Architect containerized deployments with Docker and CI/CD pipelines using GitHub Actions for seamless deployment to DigitalOcean droplets.
• Security & Compliance: Ensure all agentic workflows strictly adhere to data-protection practices, preventing LLMs from exposing sensitive policyholder or financial data outside established boundaries.
• Mentorship: Lead design and code reviews, setting quality expectations, and mentor mid-level and junior engineers on both core TypeScript and AI integration best practices.

Requirements

• 5+ years of professional software engineering experience with deep expertise in TypeScript and full-stack development.
• Proven experience designing scalable backend services, APIs, and database architectures (Prisma, PostgreSQL).
• Hands-on experience building Agentic systems or advanced LLM orchestration (e.g., using Vercel AI SDK, LangChain.js, or custom agent loops).
• Deep understanding of LLM mechanics: structured output parsing (JSON mode), function/tool calling, context window management, and token optimization.
• Experience building RAG pipelines, including working with vector databases (pgvector, Pinecone, Weaviate) and embedding models.
• Solid experience with Docker, Docker Compose, and CI/CD (GitHub Actions).
• Track record of owning features end-to-end, making sound architectural tradeoffs, and demonstrating the ability to mentor other engineers.

Nice to Have

• Experience with Bun and modern JS tooling (Biome, Vite).
• Experience with AI observability tools (e.g., LangSmith, Weave, Braintrust).
• Background building or maintaining monorepo architectures.
• Experience in insurance, fintech, or other regulated/compliance-heavy industries.
• Familiarity with prompt engineering and systematic prompt management.