Objective

• Develop a specialized AI agent that recommends the best health or life insurance plans for individuals.
• Recommendations must consider age, family status, medical history, risk tolerance, and budget.

Key features & scope

• Personalized plan comparison across product types (hospital, dental, life, disability, etc.).
• Simulation of premium vs. coverage scenarios to show trade-offs and sensitivity to inputs.
• Integration of regulatory constraints and eligibility rules so outputs respect legal requirements.
• Clear explainable outputs: “best plan + why + alternatives” for end users.

Technical stack & components

• Core stack: Python, Pandas for data processing, Streamlit for prototype UI.
• ML / NLP: HuggingFace for NER and classification models to extract and interpret user medical/history inputs.
• Rule engine: Durable Rules or a custom rule engine to encode regulatory constraints and eligibility logic.
• Target architecture: AI agents / autonomous systems approach to orchestrate model, rules, simulation and UI.

Tasks & deliverables

• Implement data ingestion and preprocessing pipelines for customer profiles and plan data.
• Train and evaluate NER/classification models (HuggingFace) to extract relevant personal and medical attributes.
• Design and integrate a rule engine to enforce eligibility and regulatory constraints into recommendations.
• Develop simulation modules that compute premium vs coverage trade-offs and generate alternative scenarios.
• Build a Streamlit prototype demonstrating the AI advisor with explainable outputs and interactive inputs.
• Deliverables: code repository, technical documentation, evaluation report, and a Streamlit demo.

Required skills & desirable experience

• Strong Python programming and data manipulation skills (Pandas).
• Experience with HuggingFace transformers for NER/classification and model evaluation.
• Familiarity with rule engines (Durable Rules or equivalent) or ability to implement rule-based systems.
• Experience building prototypes with Streamlit or similar rapid UI frameworks.
• Knowledge of insurance products (health, life, disability) or willingness to learn domain constraints.

How to apply

• Apply online via the company site: https://www.hydatis.com
• Or send your application by email to stages@hydatis.fr including CV, cover letter, and a short description of relevant projects.