Projects, Tools, and Interests

A growing part of my work is building internal tooling that makes statistical workflows more robust, more efficient, and easier to operationalize. That includes R packages, Shiny applications, and other tools that support real-world evidence generation in practice.

I am also interested in methods communication: creating concise, well-structured examples that make complex study-design ideas easier to understand without losing the underlying rigor. Together, these projects reflect the mix of tooling, methodological thinking, and practical implementation that I enjoy most.

This Shiny app is an interactive dashboard for a propensity score-based causal inference workflow. It is meant to make the workflow more tangible by surfacing the sequence of modeling, weighting, diagnostics, and interpretation in a more approachable format.

In practice, I think it is very important to separate the design and analysis phases of a causal inference workflow, with design decisions made blinded to the outcome. In this app, I intentionally combine both phases to illustrate how the full workflow can be made interactive. The app reflects the kind of tooling I enjoy building: practical, methodologically grounded, and focused on helping people engage more clearly with real analytic decisions rather than treating methods as black boxes.

This Shiny app focuses on heterogeneous treatment effects and the challenge of moving beyond average treatment effects to think more carefully about variation in response across subgroups and patient profiles.

It reflects my interest in making important causal inference ideas more interpretable and interactive, especially when the goal is to help users engage with the reasoning behind an analysis rather than just its final output.

This example walks through a clone-censor-weight style analysis in a concise, web-based format. The emphasis is less on the technology and more on clearly showing the underlying causal inference workflow and the logic behind each analytic step.

It reflects my interest in presenting complex methods in a way that stays rigorous while still being understandable to audiences working across science, analytics, and evidence-generation teams.

This example focuses on missing data and attrition in an applied causal inference setting. It is designed as a concise, web-based walkthrough that highlights why these issues matter and how they can influence interpretation if handled poorly.

Like the other methodological examples, the goal is to make the analytic reasoning more explicit rather than to showcase technology for its own sake. It reflects my broader interest in practical, readable communication of methodological choices.