PFE 1 Master AI Agent for Intelligent Insurance Routing PFE

Objective

• Conceive and implement a central "Master" AI agent that analyzes a user's personal or company profile and routes requests to the appropriate specialized insurance advisor agents (health, car, crop, business, etc.).
• Optimize routing decisions based on insurance type, urgency and contextual signals to maximize relevance and response quality.

Key Features / Responsibilities

• Implement user profile classification combining NLP on free-text inputs and structured data extraction to build a robust profile representation.
• Design and implement smart routing logic that selects the correct specialist agent(s) according to insurance category, detected urgency, and conversation context.
• Orchestrate sub-agents (specialized advisor agents), manage parallel queries, and aggregate their responses into a single, coherent recommendation for the user.
• Provide a unified user interface for recommendations (prototype via Streamlit or full UI with React) and ensure smooth API interactions (FastAPI).

Technical Stack & Deliverables

• Primary stack: Python, LangChain or Autogen, FastAPI for APIs, Redis queue for task orchestration, JSON schemas for message/agent contracts, Streamlit or React for the frontend.
• Deliverables include: working prototype of the master agent, implemented sub-agent orchestration and aggregation, JSON schemas for agent communication, deployment-ready FastAPI endpoints and basic UI showcasing end-to-end flows.
• Non-functional requirements: consider scalability (Redis queue), robustness of NLP classification, latency of routing, and clear logging/monitoring of agent orchestration flows.

How to apply / Contact

• Apply via the company site: https://www.hydatis.com
• Or send application by email to stages@hydatis.fr. Use the exact subject line: "PFE 1 - Master AI Agent for Intelligent Insurance Routing" and include a short cover note and CV.