Definitivehc Data Scientist at Definitive Healthcare focusing on applied AI and generative AI workflows. Develops and evaluates LLM, RAG, and agent-based AI solutions.
Responsibilities
Develop AI workflows: Build and test LLM, RAG, and tool-using agent workflows for assigned use cases.
Evaluate approaches: Compare LLMs, embedding models, and retrieval methods against quality, reliability, latency, and cost, using evaluation criteria and methods set by senior team members.
Maintain evaluation assets: Create and update representative test cases, benchmark datasets, and regression checks; report results and limitations.
Improve retrieval quality: Develop and tune RAG pipelines, including embeddings, retrieval, reranking, and grounding, based on evaluation results.
Investigate failures: Review incorrect answers, retrieval misses, and workflow failures; implement and measure improvements.
Support production workflows: Use experimentation, tracing, and observability tools to help monitor systems and investigate performance changes.
Collaborate on delivery: Work with engineering, product, and domain partners to clarify requirements, communicate findings, and deliver practical AI solutions.
Qualification
Education and experience3+ years of experience in data scienceTechnical and professional skillsStrong PythonUnderstanding of embeddingsAbility to design useful test casesExperience with Databricks
Required
Education and experience
Bachelor’s degree in a quantitative or technical field.
3+ years of experience in data science, machine learning, applied AI, or a related field.
Hands-on experience developing applications with LLMs and evaluating their outputs.
Technical and professional skills
Strong Python, data analysis, and problem-solving skills.
Understanding of embeddings, semantic search, retrieval, RAG and evaluation metrics.
Experience with an agent framework such as PydanticAI, LangGraph, or a similar tool.
Ability to design useful test cases, interpret evaluation results, and identify limitations.
Familiarity with cloud data or ML platforms, Git, testing, and collaborative software development.
Clear communication with technical and nontechnical colleagues.
Experience with Databricks, evaluation platforms, production deployment, fine-tuning, distributed processing, or healthcare data is beneficial but not required.