About this role
About the Role As a Data Scientist at AIG, you will serve as a technical driver for enterprise-level Generative AI initiatives, focusing on large language models and ML in production. You will collaborate across functions to design, develop, and scale AI-powered insurance solutions that manage risk and serve customers. What You'll Do
- Lead the end-to-end development and delivery of data science and generative AI solutions aligned with business needs.
- Monitor solutions in production to ensure performance, reliability, and accuracy.
- Partner with product managers, engineers, and business leaders to translate requirements into technical reality.
- Build robust evaluation frameworks to measure LLM efficacy, ground-truth data quality, and guide product roadmaps.
- Design scalable ML pipelines and RAG frameworks that integrate with enterprise data structures. What We're Looking For
- 5+ years of experience in a data science role with emphasis on NLP and ML in an agile, production-oriented environment.
- Proven delivery of multiple GenAI, analytical, or ML projects from concept to production.
- Expert-level Python proficiency and experience with PyTorch, Pandas, Scikit-learn.
- 5+ years of practical experience with open-source Large Language Models (e.g., Llama 3, Mixtral), including prompt engineering, inference optimization, and fine-tuning.
- 3+ years building Generative AI solutions, including RAG frameworks, validation pipelines, observability, and monitoring.
- Master’s degree in data science, computer science, or related quantitative field. Nice to Have
- Palantir Platform experience or certification (Foundry/AIP).
- Ontology Mastery: ability to map complex real-world entities and relationships into a digital twin framework for AI applications. Compensation & Benefits
- Total Rewards Program with comprehensive benefits focused on health, wellbeing, and financial security, plus professional development.
- Emphasis on in-person collaboration and a culture that supports ongoing learning and development.