Duration: 3 months
Employment: Full-time, Paid Internship
Compensation: Based on experience
Location: Remote
Company: FocusKPI

About the Role

FocusKPI is seeking a highly motivated PhD Data Science Intern to join our team for a three-month, full-time engagement focused on the research, development, and advancement of Media Mix Modeling (MMM) algorithms.
This is a hands-on, research-oriented role for someone with a strong foundation in statistics, econometrics, economics, or a closely related quantitative discipline who is interested in applying rigorous statistical methodology to real-world marketing and business problems.
The ideal candidate will have deep theoretical knowledge combined with practical experience developing statistical models end-to-end—from problem formulation and data preparation through model development, validation, interpretation, and production implementation.
The intern will work closely with senior data scientists and leadership to evaluate and enhance our MMM methodology, explore new modeling approaches, and translate advanced statistical techniques into scalable analytical solutions.

What You Will Do

• Research and evaluate statistical and econometric approaches for Media Mix Modeling and marketing effectiveness measurement
• Develop, test, and enhance MMM algorithms across the full modeling lifecycle
• Work with time-series, panel, and observational marketing data to develop robust models of media response and business outcomes

Explore methodologies for:

• • Media response curves and saturation effects
  • Adstock and carryover effects
  • Incrementality and causal inference
  • Channel interaction and synergies
  • Seasonality, trends, and external factors
  • Model regularization and variable selection
  • Uncertainty estimation and statistical inference
  • Bayesian and frequentist modeling approaches
• Develop model diagnostics and validation frameworks to assess model stability, predictive performance, statistical significance, and business interpretability
• Conduct simulation and experimentation to understand algorithm behavior under different data-generating conditions
• Compare alternative modeling methodologies and identify opportunities to improve model accuracy, robustness, and interpretability
• Translate research findings into production-ready algorithms and analytical workflows
• Work with real client datasets and understand the practical challenges of applying MMM to imperfect business data
• Collaborate with senior data scientists to document methodology, assumptions, limitations, and results
• Contribute to the development of next-generation MMM capabilities within FocusKPI

Required Qualifications

• PhD in Statistics, Economics, Econometrics, Applied Mathematics, Data Science, or a closely related quantitative field

Strong theoretical foundation in:

• • Statistical modeling
  • Econometrics
  • Regression and multivariate analysis
  • Time-series analysis
  • Probability and statistical inference
  • Optimization
• Strong understanding of causal inference and observational data

Demonstrated ability to develop statistical models end-to-end, including:

• • Problem formulation
  • Data preparation and feature engineering
  • Model specification
  • Estimation
  • Model diagnostics
  • Validation
  • Interpretation
  • Implementation
• Strong programming skills in Python
• Experience working with large, complex datasets
• Ability to translate mathematical and statistical concepts into practical algorithms
• Strong analytical and problem-solving skills
• Ability to work independently while collaborating closely with senior technical team members

Preferred Qualifications

• Direct experience with Media Mix Modeling (MMM)
• Experience with marketing measurement, marketing analytics, or advertising data
• Experience with Bayesian hierarchical models
• Experience with causal inference, experimentation, or uplift modeling
• Experience with time-series econometrics

Familiarity with:

• • Bayesian inference / MCMC
  • State-space models
  • Regularization
  • Constrained optimization
  • Nonlinear regression
  • Response curve estimation
  • Monte Carlo simulation
• Experience with modern statistical computing frameworks such as PyMC, Stan, NumPyro, JAX, scikit-learn, statsmodels, or equivalent
• Experience taking research concepts and converting them into reusable production code