Senior Manager, AI and Data Science

Madrigal Pharmaceuticals•Conshohocken, PA

1d•$163,000 - $200,000

About The Position

Madrigal Pharmaceuticals is advancing transformational therapies with scientific rigor and a commitment to patients. Within Medical Affairs, we turn scientific evidence — publications, real-world evidence (RWE), HEOR, and field-medical insight — into decisions that advance patient care, and we operate a growing AI platform with multiple production agents supporting that work. We are seeking a Senior Manager, AI and Data Science who brings strong technical depth, hands-on experience with agentic AI, and an ownership mindset. This individual will own modular, high-impact workstreams end-to-end, from design through production — combining agentic AI and large language model (LLM) engineering with applied machine learning on real-world evidence — working alongside the platform’s architect and Medical Affairs stakeholders. We welcome candidates from a range of scientific, medical, and engineering backgrounds. What unites strong applicants is rigor and an evidence-driven approach to building and evaluating AI systems, an ownership mindset, and the ability to collaborate across scientific, technical, and business teams. The ideal candidate is comfortable owning ambiguous problems end-to-end.

Requirements

A demonstrated record of scientific rigor: either a Ph.D. in a quantitative or medical-adjacent field with first-author publications, or a strong engineering background paired with peer-reviewed research at top venues (for example, NeurIPS, ICML, ICLR, or ML4H). The specific credential matters less than the demonstrated rigor and record of shipping.
7+ years of applied machine learning or data science, including 3+ years shipping ML or AI systems to production.
1+ year of hands-on agentic AI development using modern frameworks (for example, LangGraph, deepagents, or LangSmith).
Strong RAG and retrieval engineering, including vector and hybrid search, embeddings, and reranking, with the ability to evaluate groundedness and citation accuracy.
Applied ML on real-world or patient-level data (claims, EHR, registry), including feature engineering.
Experience with production reliability, data access and authentication, and handling sensitive (PHI) data.
Strong communication skills and the ability to collaborate across scientific, technical, and business teams.
Experience in Healthcare or Life Sciences.

Nice To Haves

Production experience with deepagents, LangGraph, and LangSmith; MCP tools; Azure; and Databricks Unity Catalog, Microsoft Fabric, or OneLake.
Proven track record of deploying agents to production at scale, including LLM evaluation and observability.
Background in real-world evidence, epidemiology, or HEOR analytics.
Medical Affairs experience.

Responsibilities

Build and operate production AI agents and orchestration workflows using modern agentic frameworks such as LangGraph and deepagents.
Design modular, reusable components for reasoning, retrieval, tool integration (including MCP), and workflow automation.
Partner with the Agentic UI team to deliver seamless, end-to-end products.
Own and optimize the retrieval stack, including vector and hybrid search, embeddings, chunking, and reranking.
Build evaluation and regression safeguards for LLM outputs, with particular attention to citation faithfulness and groundedness, which are essential in Medical Affairs.
Design and deploy machine learning on real-world evidence (claims, EHR, registry), HEOR, and field-medical data to surface Medical Affairs insights.
Apply embeddings, predictive models, and statistical methods where they are the right fit for the problem.
Automate and scale data-ingestion pipelines into AI/ML-ready data on platforms such as Databricks.
Own deployment, monitoring, and observability (for example, using LangSmith) for the systems you ship.
Translate complex AI concepts into clear, actionable insights for technical and non-technical Medical Affairs partners.
Align AI initiatives with enterprise governance, privacy, and PHI and compliance expectations.