Bluespine-Senior Data Scientist

Description

Bluespine is an innovative new startup in the health-IT domain. By employing cutting-edge technologies, Bluespine is developing an engine that detects errors in medical billing, which causes billions of dollars in losses across the entire industry. Bluespine can offer personalized precision by tailoring assessments to each unique medical claim, considering the relevant provider, payer, and plan, and ensuring unparalleled accuracy.

We are looking for an expert data scientist, who is familiar and experienced with applying machine learning and deep learning in a commercial setting. This role requires proven hands-on experience training and optimizing models, building production-ready inference pipelines, performing exploratory data analysis & enrichment, and validating models for issues like bias, overfitting, and concept drift.

Responsibilities

• Lead all aspects of building our core AI/ML models and assets, from inception to productization.
• Develop models for data extraction from unstructured documents, logical reasoning over policy documents, ontology extraction/population and rules generation.
• Develop training and evaluation pipelines that incorporate human feedback.
• Collaborate closely with our R&D team as well as domain experts in the fields of medical coding and billing.

Requirements

Requirements

• Experience in building production-ready deep-learning NLP models or LLMs with RAG.
• Relevant industry experience with demonstrable impact in a commercial setting.
• Significant practical usage of modern ML areas (NLP, Transformers, LLMs, NER, GANs, Reinforcement Learning, etc.)
• Strong analytical skills and ability to approach tasks in a scientific manner.
• Data science skills, including modeling and evaluation.
• Independent, Organized, and with excellent communication skills.

Advantages

• Experience in ontology mining/rules inference systems.
• Prior background in health insurance tech / medical coding/billing.
• Degree in computer science or equivalent.