Postdoctoral Fellow – Agentic AI Solutions and SciML Full-time NEW

Postdoctoral Fellow, Agentic AI Solutions and SciML

At Lilly, the work is demanding because patients are waiting. We unite caring with discovery to help make life better for people around the world, knowing that every decision, every detail, and every day matters. Headquartered in Indianapolis, Indiana, our over 50,000 employees around the globe take on complex challenges to discover and deliver life-changing medicines, strengthen how health is understood and managed, and support the communities we serve. This is hard, urgent, selfless work—but it's work worth doing. If you're driven by purpose and ready to bring your best to work that truly matters for patients, we invite you to join us.

Delivery, Devices, and Connected Solutions (DDCS) sits within Eli Lilly's Product Research & Development organization. We are a diverse team of scientists and engineers responsible for discovering, designing, and developing patient-centric drug delivery solutions across a broad range of modalities — from injection devices to novel routes of administration and nanomedicines. DDCS drives the drug delivery innovation agenda across early and late development to meet the needs of an expanding portfolio that spans small molecules, biologics, and nucleic acid therapeutics.

DDCS is organized around a matrix model with strong disciplinary and functional horizontals supporting innovation and commercialization verticals. Our vision is to get our medicines to more patients faster by accelerating reach and scale, guided by three strategic pillars: Delivery Systems, Robust & Sustainable, and Patient Experience + Outcomes.

The Digital Transformation and Data Science team within DDCS serves as a key foundation for DDCS's digital transformation efforts. The team helps make data work more effectively for the organization by turning information into faster insights, stronger decision-making, improved ways of working, and practical AI solutions embedded in everyday DDCS workflows.

The Postdoctoral Fellow, Agentic AI Solutions and SciML will partner with DDCS business functions to help translate machine learning, statistics, scientific computing, and AI concepts into practical tools that improve speed, productivity, and impact across the organization. Operating within the Digital Transformation and Data Science team, this role will assist in the design, build, and deployment of AI-enabled workflows, agentic scientific systems, knowledge extraction tools, and scientific ML capabilities into actionable decisions.

Key Responsibilities:

AI Solutions Engineering & Practical Tool Delivery

• Partner with business functions across DDCS to identify, prioritize, and scope high-value opportunities where AI, machine learning, and automation can improve speed, productivity, insight generation, and decision quality.
• Translate stakeholder needs into practical AI tools, technical designs, acceptance criteria, and delivery plans that fit real scientific, engineering, and operational workflows.
• Develop AI-enabled applications, services, and workflows that integrate models, data sources, document collections, and user-facing interfaces for decision support and workflow automation.

Agentic Scientific AI Systems & Knowledge Extraction

• Create reusable scientific agent skills, task harnesses, validators, run ledgers, and reproducibility controls that allow AI agents to execute diverse, long-running tasks reliably.
• Build agentic knowledge extraction and question-answering systems for structured and unstructured technical content, including PDFs, Word documents, handwritten notes, design histories, experimental records, and regulatory-relevant evidence.
• Design evaluation, monitoring, guardrails, and human-in-the-loop escalation patterns so agentic outputs are auditable, traceable, and appropriate for high-consequence technical decisions.
• Apply knowledge graphs, data ontologies, and structured knowledge representation where they improve retrieval, traceability, and reuse.

Data Strategy, Decision Support & Workflow Transformation

• Contribute to DDCS data and AI strategy by identifying reusable patterns, data needs, platform capabilities, and solution architectures that support digital transformation at scale.
• Turn information from experiments, simulations, development documents, and business processes into faster insights, stronger judgment, and improved ways of working across innovation and commercialization efforts.
• Communicate model predictions, evidence, assumptions, limitations, uncertainty, and recommended actions through clear visualizations, decision-support outputs, and quantitative business cases that influence solution adoption, workflow redesign, platform investments, and portfolio priorities.

Reliability, Validation, MLOps & Responsible AI

• Champion software engineering best practices including version control, automated testing, CI/CD, containers, documentation, reproducibility, observability, and fit-for-purpose MLOps/agent-ops practices.
• Develop validation, monitoring, documentation, and model-risk approaches aligned with intended use, responsible AI principles, GxP awareness, and regulatory expectations where applicable.
• Leverage cloud infrastructure (and HPC/GPU resources where needed) to develop, test, deploy, and scale agentic workflows, document intelligence systems, and analytics applications.

Cross-Functional Collaboration & Scientific Translation

• Partner across the DDCS matrix with drug delivery scientists, device engineers, formulation scientists, data scientists, AI application engineers, quality, clinical, regulatory, and business stakeholders.
• Identify and prioritize high-impact opportunities where AI solutions, scientific ML, agentic workflows, or knowledge extraction can reduce development time, improve productivity, or mitigate technical and business risks.
• Translate complex analytical and AI findings into clear narratives and quantitative business cases that influence solution adoption, workflow redesign, platform investments, and portfolio priorities.

Capability Building, External Leadership & Mentorship

• Advance the DDCS technology roadmap for practical AI tools, document intelligence, agentic workflows, and reusable knowledge systems.
• Share methods, reference patterns, and lessons learned that help DDCS embed data and AI into everyday work across scientific and business functions.
• Mentor team members and partners on reliable agentic systems, responsible AI, and rigorous communication of model assumptions, uncertainty, and decision impact.
• Stay current with the fast-moving agentic AI and LLM landscape and bring new tools, frameworks, and techniques into DDCS's practice where they add real value.

Basic Qualifications

• Earned PhD with up to 5 years of experience in Computational/Computer Science, Machine Learning, Artificial Intelligence, Engineering, or a related quantitative field (or equivalent experience)
• Applied technical work building AI or machine learning solutions in a programming language such as Python/R, with working knowledge of the ecosystem (NumPy, pandas, PyTorch, scikit-learn, or related).
• Strategic thinking, problem framing, and translating ambiguous business or scientific needs into tractable AI, modeling, or computational workflows.
• Demonstrated ability to frame ambiguous business or scientific needs as tractable AI, modeling, or computational workflows.
• Skill in communicating technical recommendations with clearly stated assumptions, uncertainty, and limitations, to scientific, engineering, and business audiences.

Additional Preferences:

• Experience applying AI/ML to healthcare, pharmaceutical, or life-sciences problems (prior biology or life-sciences background not required).
• Strong SQL and relational data modeling, with comfort turning large, messy, unstructured, or incomplete data into reliable, decision-ready output.
• Experience with cloud platforms and solid engineering practice: Git, containers, CI/CD, and experiment or run tracking. Comfort with GPU and HPC environments is a plus.
• Experience with knowledge graphs, ontologies, or structured knowledge representation for technical content.
• Evidence of contribution to significant work, ideally through publications at relevant ML/AI/NLP venues (NeurIPS, ICML, ICLR, ACL, EMNLP) or comparable open-source or applied contributions.
• Fluency with agent frameworks and orchestration (LangGraph, AutoGen, CrewAI, or equivalent) and the primitives underneath them: planner/executor splits, hand-offs, escalation logic, and state management across multi-step or multi-session workflows. Knowing why they fail, not just how to call them.
• End-to-end RAG design over messy technical documents: parsing and layout extraction from PDFs, scans, and tables; chunking strategy; hybrid search; reranking; embedding models; and vector stores (pgvector, Pinecone, Weaviate, or similar).
• LLM engineering judgment: context design, tool/function calling (MCP or comparable standards), structured output design at scale, and knowing when to fine-tune versus retrieve versus prompt.
• Evals engineering: golden datasets and benchmarks, automated regression suites, and failure-mode tracking. Evidence of a trustworthy agent to deploy.
• LLMOps in production, treating cost, latency, and reliability as engineering constraints with the monitoring to match.
• Guardrail and safety design for autonomous systems: approval gates, rollback logic, hallucination and drift detection, and model-risk thinking for