PhD Intern - Data Scientist & Generative AI in Femcare

Procter & Gamble•Cincinnati, OH

1d•Onsite

About The Position

Research and Development (R&D) at Procter & Gamble (P&G) includes a diverse group of roles that contribute to the innovation and development of our products. This internship is designed for individuals working towards a PhD who are developing proficiency in their field. As an intern or co-op in management, you will have the opportunity to learn from experienced professionals in a supportive environment. This is a 12-week paid internship, designed to provide you with a solid foundation for future career growth. The internship will take place from May (potentially early June) to August of 2027. P&G has an opportunity for a PhD intern to work in our Feminine Care business working in the space of Data Scientist & Gen AI. This role leverages current Gen AI technology, multi-discipline knowledge and R&D product innovation cycle, especially by leveraging data and digital approaches. The candidate will create Gen AI tool as co-pilot to guide and influence technical project direction and enable fast problem solving by leveraging collective knowledge in our data & modeling eco-system. The candidate should demonstrate strong scientific/engineering skills through advance model development and application development. You will investigate how active learning and sequential experimentation can accelerate learning in R&D. Using historical consumer study and simulation datasets, the project will evaluate whether uncertainty estimates can identify the most informative next study, measurement, or simulation to perform. Rather than exploring the design space uniformly, the goal is to prioritize experiments that are expected to provide the greatest reduction in uncertainty and the highest learning value. This work will treat consumer studies and first-principles simulations as two applications of the same underlying problem: how to allocate experimentation resources more efficiently.

Requirements

Working towards a PhD in Computer Science, Machine Learning, Statistics, Mechanical Engineering, Computational Physics, Computational Engineering, Applied Mathematics, or Data Science.
Active learning / sequential experimental design — central to recommending the next most informative consumer study or simulation.
Probabilistic machine learning / Bayesian statistics — needed to represent prediction uncertainty and update confidence as evidence arrives.
Surrogate modeling — needed for both consumer-response prediction and simulation-response approximation.
Python — required for model development, simulation, and analysis; PyTorch, scikit-learn, BoTorch, GPyTorch, or similar tools would be useful.
Experimental design and model evaluation — needed to compare active learning policies against baselines and define success metrics.
Available to work a 12-week internship from May/early June to August in the summer of 2027, and at least 3 days/week onsite at Winton Hill Business Center.

Nice To Haves

Experience with entrepreneurial/startups and/or industrial experience are strongly preferred.
Has strong ownership & self-leadership skills to understand the business objective and technical problem to be solved.
Has experience collaborating with multi-functional teams.
Has the ability to recognize opportunities for capabilities or technologies that can be reapplied across programs.
Can independently work project priorities, multi-task, and handle a wide variety of complex, non-routine/routine tasks in a fast-paced environment.
Bayesian optimization
Multi-armed bandits or reinforcement learning
Uncertainty quantification
Gaussian processes, evidential models, Bayesian neural networks, or ensembles
Scientific computing
CFD / FEA familiarity, if using first-principles simulation examples
Generative AI / AI-agent prototyping, but only as a communication layer, not the core research contribution

Responsibilities

Using historical consumer study and simulation datasets to evaluate uncertainty
Identify next study, simulation, and measurement based on reduction of uncertainty
Optimize experimental resources
Create a prototype work process to demonstrate success