Senior Biostatistician — R / Mixed-Effects Modeling / Predictive Analytics

Employment Type: Full time, independent contractor
Work Model: Remote
Work Hours: US work hours

Role Overview

We’re hiring a senior biostatistician to lead our statistical modeling and predictive analytics work.

This role owns the modeling architecture: designing and fitting longitudinal mixed-effects models, extracting coefficients and variance-covariance matrices, building dose-response curves, and powering the prediction engines behind our client-facing dashboards.

You need to be able to do this work from day one. We’re not looking for someone to train up — we need someone who has built these models before and can hit the ground running.

This is a fully remote position.

Core Responsibilities

Statistical Modeling & Predictive Analytics

Design, specify, and fit mixed-effects models (linear and generalized) for longitudinal observational data
Extract beta coefficients and variance-covariance matrices from fitted models and export them for use in downstream prediction engines (JSON format → JavaScript calculators)
Build and embed core calculator logic — the math that powers client-facing predictive dashboards
Construct dose-response curves across product types, formulations, and outcome measures
Handle model convergence issues, optimizer tuning, random effects specification, and model selection independently
Cross-cohort and cross-product comparative analyses (global models, interaction effects, brand-level vs. cohort-level comparisons)
Develop novel analytical approaches for new study types and research questions

Industry-Level Analysis

Build cohort-level industry reports — aggregate models across multiple products and brands within a study cohort
Design cross-cohort comparative methodology as the dataset grows
Cannabinoid predictability analysis: formulation × product type × dose → effect prediction
Statistical architecture for new outcome domains (pain, oncology, non-cannabinoid substances)

Methodology & Leadership (10%)

Own statistical methodology decisions for the company
Validate and review analytical outputs
Collaborate with the dev team on model-to-dashboard pipeline (you provide the statistical outputs, they build the software)
Mentor the data analyst on advanced statistical concepts
Document methodology for reproducibility and regulatory defensibility

Required Skills

Non-Negotiable

R (advanced, primary language) — must be fluent. This is not a Python shop.
Mixed-effects models — lme4 package (lmer, glmer). Must understand random effects specification, convergence diagnostics, optimizer tuning (bobyqa, nelder-mead, nAGQ), model comparison, and cross-optimizer validation (allFit()).
Coefficient extraction — ability to extract fixed-effect coefficients, variance-covariance matrices, and random-effect parameters from fitted models and structure them for use outside R (JSON, CSV, matrices).
Longitudinal data analysis — repeated measures, within-subject designs, time-varying covariates, handling attrition and missing data
Generalized linear mixed models (GLMMs) — binomial, Poisson, ordinal outcomes
Dose-response modeling — experience building dose-response or exposure-response curves
Statistical thinking at a senior level — ability to design analyses from scratch for novel research questions, select appropriate models, justify decisions, and interpret results in context. Not just running code someone else wrote.

Strongly Preferred

Real-World Evidence (RWE) — experience translating real-world data into defensible evidence in a health or consumer product context
Experience with observational (non-randomized) study designs, including causal inference and propensity score methods
Predictive model deployment — exporting model parameters for use in applications, not just in R. Experience building scoring algorithms, personalized prediction engines, or dosing calculators from statistical models
Automated R/Quarto pipelines — building workflows that ingest longitudinal datasets and auto-generate reports, not just one-off analyses
ggpredict() / effects / marginaleffects packages for marginal predictions
Cross-study or meta-analytic methodology (cross-cohort comparison, pooled analyses)
Health/biomedical/clinical research background
Experience with survey data, Likert-scale outcomes, or validated psychometric instruments (WHO-5, DASS-21, SF-36, etc.)

Nice to Have

Power analysis and study design
Python (secondary)
Bayesian methods (brms, Stan)
Experience with IRB-governed research
Cannabis/cannabinoid research experience
JSON/data serialization for model export
Shiny or Plotly for interactive statistical visualization

Work Environment

Fully remote
Full-time (160 hrs/month)
Reports to COO. Works alongside a full-time data analyst.
AI-assisted workflow — we use AI tools extensively for coding and analysis. You should be comfortable leveraging AI tools and collaborating with AI-powered statistical review.
Small team, startup pace — significant autonomy and direct impact on the company's analytical capabilities

What Success Looks Like (Subject to Change)

Month 1

Run our existing mixed-effects model pipeline on a new study dataset. Extract coefficients and vcov matrices. Validate outputs against existing delivered reports. Demonstrate you can do the core modeling work independently.

Month 2–3

Build a cohort-level industry report from scratch — global models, per-brand models, interaction effects, cross-product analysis. Own the dose-response curve pipeline. Extend the coefficient extraction pipeline to new outcome variables.

Month 4–6

Design and execute novel analytical approaches for new study types and research questions. Own the full statistical architecture of the company. Lead methodology decisions.

Ideal Candidate Profile

Must-Have

R as primary language — listed as primary, not secondary. Multiple years. Should be the tool they think in.
Mixed-effects / multilevel models — explicit mention of lme4, lmer, glmer, or "mixed-effects" / "multilevel" / "hierarchical" models. Not just "regression."
Coefficient extraction or model deployment — any evidence of taking model outputs outside R — exporting coefficients, building prediction tools, deploying models into applications, or creating calculators/dashboards from model parameters.
Longitudinal / repeated-measures data — experience with panel data, repeated measures, or time-series in a health or behavioral context.
Graduate degree (PhD preferred) — PhD in biostatistics, statistics, epidemiology, or quantitative health science. Master’s acceptable with 5+ years applied experience.
Senior-level independence — evidence of owning methodology decisions, not just executing someone else’s analysis plan.

Strong Differentiators

Dose-response modeling — explicit mention of dose-response, exposure-response, or pharmacological/pharmacokinetic modeling.
Health/biomedical domain depth — published research, clinical trials, observational studies, epidemiological work.
GLMMs beyond linear — binomial/logistic mixed models, ordinal regression, count data models.
Cross-study / meta-analytic work — experience comparing across studies, cohorts, or populations.
Predictive analytics — building prediction engines, not just fitting models.
Consulting / applied work — experience producing work for clients or stakeholders, not purely academic.

Senior Biostatistician - 18162