Researcher I - AI Researcher

Requirements: Required Skill Set: The qualified candidate will have a combination of the following: Experience designing and implementing LLM and agentic AI systems, including RAG, tool/function calling, multi-agent orchestration, prompt and workflow design, evidence-grounded response validation, hallucination mitigation, fallback routing, and failure analysis. Experience developing trustworthy, audit-focused AI workflows with audit logging, traceable workflow design, execution monitoring, source-linked evidence artifacts, run metadata, model risk documentation, and reproducible technical documentation. Knowledge of model evaluation and validation practices, including data quality checks, robustness testing, drift or failure monitoring, human review controls, and documentation of system assumptions, dependencies, limitations, and risk mitigations. Ability to translate reviewer and scientific user needs into practical AI-assisted workflows and user-facing prototypes while maintaining appropriate controls for non-public information and high-stakes regulatory use. T asks and responsibilities: Design, develop, test, and document AI solutions to support drug submission review workflows, including document triage, evidence extraction, literature and submission information retrieval, question answering, summarization, and generation of traceable review artifacts. Develop and integrate human-in-the-loop review features for AI solutions, including reviewer feedback capture, source verification, confidence or risk flagging, issue escalation, editable outputs, audit trails, and mechanisms for scientific review and approval before downstream use. Build pipelines to ingest, parse, structure, and analyze regulatory, scientific, and clinical documents and data sources, including PDFs, tables, figures, metadata, and structured or unstructured text, while preserving traceability to source evidence. Establish evaluation and validation protocols to assess AI output accuracy, reproducibility, robustness, source attribution, hallucination risk, tool-use reliability, and operational limitations; identify unexpected results and provide potential causes and mitigation strategies. Implement logging, monitoring, and quality assurance controls for AI workflows, including tool/function call validation, data quality checks, execution monitoring, failure reporting, token or resource usage tracking, and reproducible artifact generation. Prepare technical documentation, workflow descriptions, risk and limitation summaries, user guidance, progress reports, and other deliverables needed to support transparent development, human oversight, and future maintenance of AI-enabled review tools.

Education and Experience: These projects require scientists with a bachelor's, master's degree or Ph.D. in data science, computer science, biomedical informatics, engineering, statistics, or a related quantitative field, and experience developing, validating, documenting, and deploying AI/ML, large language model (LLM), retrieval-augmented generation (RAG), and/or agentic workflow systems in biomedical, clinical, regulatory, or other high-stakes settings. Experience with auditability, traceability, human-in-the-loop review, responsible AI controls, and production software development is preferred. PPO/HMO Health Plan (includes medical, dental, and vision)