2027 Future Talent Program – Translational Sciences and Outsourcing – Intern
MSD
Summary
The Future Talent Program features internships that last up to 12 weeks and will include one or more projects within the Research and Development Division. We are seeking a highly motivated intern student to join the Translational Sciences and Outsourcing (TSO) Data Science team within our Pharmacokinetics, Dynamics, Metabolism, and Bioanalytics (PDMB) department.
Job description
Project Overview
The intern will benchmark existing AI/ML methods for predicting pharmacokinetic concentration-time profiles using a large-scale internal dataset. The project will determine how model choice, data-splitting strategy, chemical similarity, compound novelty, and variation across data sources affect predictive performance and prospective generalization. The work is scoped to produce a rigorous, reproducible benchmark study and, subject to scientific results and internal review, a paper suitable for submission to a peer-reviewed journal or scientific conference.
Primary Responsibilities
- Analyze large-scale pharmacokinetic and molecular datasets and prepare them for reproducible modeling and evaluation.
- Implement and benchmark classical machine learning baselines, molecular-structure-based models, and representative time-series or mechanism-informed approaches for pharmacokinetic curve prediction.
- Design and compare random, scaffold, time-based, and chemical-similarity-aware evaluation settings, and quantify performance across compound novelty levels to identify data leakage and limitations in prospective generalization.
- Collaborate with scientists to interpret results and ensure that benchmark conclusions are scientifically meaningful.
- Document methods and results and contribute to a draft manuscript or technical report.
- Present project outcomes at the end-of-summer intern symposium.
Learning Opportunities
- Pharmacokinetics and its applications in drug discovery.
- Machine learning methods for concentration-time curve prediction.
- Molecular representations, similarity analysis, chemical clustering, and rigorous benchmark design.
- Reproducible research and cross-functional collaboration within a pharmaceutical research organization.
- Scientific writing, communication, and presentation skills.
Expected Deliverables
- A reproducible evaluation workflow with documented datasets, train/test splits, model configurations, metrics, and visualizations.
- A final presentation summarizing project objectives, methods, results, scientific conclusions, limitations, and recommended next steps.
- A draft manuscript describing the scientific question, benchmark design, comparative results, chemical-novelty analyses, limitations, and conclusions, prepared with the project team and subject to internal publication review.
Requirements
- PhD student in computational chemistry, cheminformatics, machine learning, computer science, computational biology, statistics, or a related field
- Strong record of publication in top-tier, peer-reviewed scientific journals or machine learning conferences (e.g., ICML, ICLR, NeurIPS)
- Strong Python programming skills
- Experience using machine learning or deep learning frameworks such as PyTorch or similar tools
- Experience with model development and evaluation
- Ability to work with complex scientific datasets and build reproducible analysis pipelines
- Understanding of how data leakage and evaluation design affect model generalization
- Experience designing comparative studies and building reproducible research workflows
- Strong communication skills and ability to summarize technical results clearly
- Preferred: Experience with pharmacokinetics, pharmacometrics, ADMET, or drug discovery
- Preferred: Familiarity with PK concepts such as concentration–time curves, AUC, Cmax, clearance, half-life, compartment models, and NCA
- Preferred: Experience with molecular representations and cheminformatics tools including SMILES, molecular fingerprints, graph neural networks, Chemprop, Uni-Mol, or RDKit
- Preferred: Familiarity with molecular property prediction, QSAR, or ADME modeling
- Preferred: Experience with molecular similarity analysis such as Tanimoto similarity, scaffold splitting, chemical series clustering, or out-of-distribution evaluation
- Preferred: Experience with time-series modeling, ODE models, neural ODEs, or physics-informed machine learning
Company overview
MSD, known as Merck Sharp & Dohme outside the United States and Canada, is a global research-intensive biopharmaceutical company headquartered in Rahway, New Jersey. Founded in 1891, the company has established itself as one of the world's leading pharmaceutical corporations with a rich history spanning over 130 years. MSD operates as the international division of Merck & Co., Inc., focusing on the discovery, development, manufacturing, and marketing of innovative medicines, vaccines, biologic therapies, and animal health products. The company's extensive portfolio includes groundbreaking treatments in oncology, infectious diseases, cardiology, diabetes, and vaccines, with notable products such as Keytruda (pembrolizumab) for cancer immunotherapy, Gardasil for HPV prevention, and Januvia for diabetes management. MSD maintains a strong commitment to research and development, investing significantly in scientific innovation to address unmet medical needs worldwide. With approximately 75,000 employees globally, the company operates through four main segments: pharmaceuticals, animal health, healthcare services, and strategic alliances. MSD's mission centers on using leading-edge science to save and improve lives around the world, guided by the principle that medicine is for people, not profits.