About the role
Role: GenAI / GCP AI Engineer
Location: Hartford, CT
Fulltime
Role summary
Build, test and deploy production GenAI solutions on Google Cloud for insurance clients: document summarization, intelligent intake, underwriting and claims assistants, and agentic workflows. You turn architecture into working code and own features from prototype to production.
Key responsibilities
Build GenAI applications with Gemini models on Gemini Enterprise Agent Platform (Vertex AI).
Design and tune prompts, structured outputs and function calling for insurance documents (submissions, ACORD forms, COIs, loss runs, policies).
Build RAG pipelines: chunking, embeddings, Vector Search / Vertex AI Search, grounding and citations.
Build agents with the Agent Development Kit (ADK) and deploy them on Agent Engine; integrate tools and APIs.
Extract and classify documents with Document AI and Gemini multimodal input.
Ship services on Cloud Run or GKE, exposed through Apigee, with CI/CD in Cloud Build.
Move and store data with BigQuery, Cloud Storage, Pub/Sub and Dataflow.
Write evaluation sets and run quality, latency and cost checks (Gen AI evaluation service, golden datasets).
Apply guardrails: PII masking with Sensitive Data Protection (DLP), safety filters, IAM least privilege.
Work in Agile sprints with QA, UI and data teams; write clean, tested, documented code.
Required skills
Strong Python (FastAPI or Flask, async, pytest).
Hands-on Gemini API / Vertex AI SDK or Google Gen AI SDK.
RAG design and at least one vector store (Vector Search, AlloyDB pgvector, BigQuery vector search).
LLM frameworks: ADK, LangChain or LangGraph.
GCP core: Cloud Run, Cloud Functions, Cloud Storage, BigQuery, IAM, Secret Manager.
REST API design, Git, Docker, CI/CD.
Good written and verbal communication with onsite and client teams.
Nice to have
Insurance domain (P&C, E&S, specialty, workers' comp) or financial services.
Document AI custom extractors; OCR on scanned or handwritten forms.
Agent2Agent (A2A) protocol or Model Context Protocol (MCP) tool integration.
Model tuning (supervised fine-tuning on Gemini) and model monitoring.
Terraform; Looker or Looker Studio dashboards.
Certifications (preferred)
Google Cloud Professional Machine Learning Engineer
Google Cloud Associate Cloud Engineer or Professional Cloud Developer
Google Cloud Generative AI Leader (good to have)"
Requirements
- Strong Python (FastAPI or Flask, async, pytest).