Ruwan C. Karunanayaka, Ph.D., P.Stat.
  • Consulting
  • Projects
  • Research
  • Teaching
  • Blog
  • ORCID

Statistical Consulting

Statistics you can defend.

I’m an accredited professional statistician (P.Stat.) and Associate Professor who helps researchers, teams, and organizations design studies properly, model data honestly, and deliver results that hold up to scrutiny — from the first power calculation to the final dashboard.

Work with me See the tools I build

What clients get

A defensible answer, not just a number: the right design for the question, methods chosen for the data rather than habit, uncertainty stated plainly, and a writeup or interactive tool your stakeholders can actually use.

study design   DOE   mixed models   reliability   AI evaluation   R / Shiny

Credentials

Grounded in peer-reviewed work.

Training & position

PhD in Statistics (Simon Fraser University, 2018). Associate Professor, Department of Mathematics & Statistics, University of the Fraser Valley.

Published methods

Peer-reviewed research in experimental design and applied statistics, with methods independently cited across textile engineering, food science, hydrology, and mining research.

Accreditation

P.Stat. (Professional Statistician), accredited by the Statistical Society of Canada. SSC member since 2017.

What I help with

Three services, one standard.

Study & experiment design

Sample size and power, randomization, factorial and fractional-factorial designs, including the baseline parameterization — so the data you collect can actually answer your question.

Modelling & analysis

Mixed-effects and generalized linear models, measurement reliability, calibration and validation. Documented, reproducible, reviewable.

Tools & dashboards

Interactive R / Shiny applications that put your model in front of the people who need to act on it — see live examples.

Statistics for the AI era

Your AI is only as good as its evaluation.

Benchmark scores drift with the test harness, top-of-leaderboard gaps sit inside statistical noise, and most evaluation pipelines are never checked for reliability. Before an AI number drives a decision — which model to buy, whether a system is ready to ship — someone should ask whether the measurement itself can be trusted. That’s measurement science, and it’s what I do.

Evaluation design & audit

Design AI evaluations that answer the question — proper sampling, contamination control, uncertainty on every comparison — or audit the one you have.

Reliability of judgments

Human raters and LLM-as-judge pipelines are measurement instruments. I quantify their dependability and tell you how many judgments a trustworthy verdict needs.

Experiments that respect your budget

When every evaluation run costs money, designed experiments get you more answer per dollar than one-at-a-time testing of prompts, configurations, and models.

AI services in detail →

Beyond consulting

Research, software, and teaching.

Projects

Deployed forecasting apps and open-source software — working demonstrations of the tools I build. Explore →

Research

Published work in experimental design and applied statistics, with an active research program in measurement reliability. Publications →

Blog

Notes on statistical methodology — how to design, measure, and trust the numbers. All posts →

© 2026 Ruwan C. Karunanayaka, Ph.D., P.Stat. · Associate Professor, University of the Fraser Valley

 
  • ruwan.karunanayaka@ufv.ca

  • GitHub

  • ORCID