Note: The job is a remote job and is open to candidates in USA. ISF, Inc. is seeking a Quantitative Reviewer for an applied public health research project. The role involves providing quality assurance and advisory support by reviewing predictive model specifications, statistical outputs, and ensuring accuracy and consistency in a state-level decision-support system for a government agency.
Responsibilities
- Read and evaluate the statistical methods described in the project’s technical framework document to assess whether the model specification is internally consistent, the assumptions are appropriate, and the described approach is correctly implemented
- Flag methodological concerns, specification errors, or inconsistencies between the described methods and standard practice in Bayesian spatial modeling or public health surveillance
- Check that model output values are plausible, internally consistent, and correctly reported in tables and figures, including latent population estimates, detection probabilities, geographic risk scores, treatment effect estimates, convergence diagnostics, and scenario projections
- Verify that numbers cited in the technical document match the underlying model output files, and that calculations (rates, percentages, aggregations, credible intervals) are arithmetically correct
- Review sections of the technical deliverable as they are updated to confirm that numerical values, statistical summaries, table entries, and methodological descriptions accurately represent the underlying analytical work
- Identify any places where results are mischaracterized, ambiguously described, or where the documentation does not match model outputs
- Provide written review comments for the lead scientist to address
- Following data refreshes of the project’s Azure-hosted decision-support tool, spot-check displayed values (county-level counts, rates, projections, and KPI figures) against source model output files to confirm the tool is correctly reflecting updated results
- Provide time range estimates for requested QA outputs to the Principal within 24 hours of tasking
- Provide weekly updates regarding work completed to the Principal
- Provide objective feedback related to the predictive model and its outputs to the Principal, advising on future scoping with the client as appropriate
Skills
- Education: Doctorate (Ph.D.) in biostatistics, statistics, health data science, epidemiology, or closely related quantitative field
- Experience translating highly technical concepts with simplicity and accuracy to non-specialist audiences
- Experience accurately estimating the time required to complete tasks
- Experience advising leadership regarding technical processes, outputs, and hours required to complete scope
- Must be proficient in R, Azure, Azure Databricks, Claude Code
- Must currently hold, or have the ability to obtain, CITI certification to access restricted data
- Strong candidates will bring a doctoral credential alongside meaningful experience delivering technical work in externally accountable contexts, whether through consulting, applied research, government advisory work, or a combination
- Ability to operate with professionalism and clarity in a client-service environment: translating complex findings for non-specialist audiences, managing your own time and scope reliably, and engaging with institutional clients
- Comfort operating as an independent contractor within a structured engagement: estimating and tracking your own hours, meeting deadlines with appropriate autonomy, and communicating proactively with a manager when scope or schedule questions arise
- Solid working knowledge of Bayesian hierarchical modeling, including familiarity with MCMC estimation, estimating priors, convergence diagnostics (R-hat, ESS, trace plots), and posterior predictive checks
- Familiarity with spatial random effects models, preferably with experience in geospatial modeling using tools such as ArcGIS and R packages inclusive of Conditional Autoregressive (CAR) or Intrinsic CAR (ICAR) structures used in disease mapping and small-area estimation
- Understanding of abundance models or N-mixture frameworks that estimate latent populations from multiple partial observation processes, or equivalent experience with latent variable models in a public health context
- Ability to verify that reported statistics (regression coefficients, credible intervals, cross-validated performance metrics, feature importance rankings) are correctly calculated and appropriately interpreted
- Familiarity with penalized regression methods (elastic net, lasso) and cross-validation approaches, sufficient to evaluate whether a risk score construction methodology is sound
- Comfort with gradient-boosted tree methods and SHAP-based feature importance at a conceptual and evaluative level
- Strong numeracy: the ability to catch arithmetic errors, implausible values, and internal inconsistencies in tables of model results is a core requirement
- Sufficient familiarity with public health data and disease surveillance to assess whether model outputs (prevalence estimates, detection probabilities, county-level risk rankings) are epidemiologically plausible and appropriately caveated
- Understanding of small-area estimation challenges, suppression and interval censoring in public health data, and the ecological inference limitations relevant to census tract-level models
- Deep comfort with coding is important for this role. The work involves reading, running, and evaluating R scripts across a complex multi-source analytical pipeline, and the ability to move through code confidently is central to the QA function
- Strong experience with highly organized data storage practices and pipeline development. The project involves a structured Azure-based data environment with a layered medallion architecture; candidates should be comfortable working within and contributing to organized, well-documented data pipelines rather than ad hoc analytical workflows
- Ability to assess a defined scope of work and offer a reasonable hour estimate before beginning
- Comfort surfacing scope questions and clarifying tasks early
- Experience tracking and reporting hours on consulting or contract work
- This role reports directly to the Principal. Technical collaboration with data scientists is expected; however, tasking, deadlines, and client engagement will be directed by the Principal
- Client interaction may be requested by the Principal, including deliverable walkthroughs and discussions related to the model framework, model inputs, model outputs, etc. The successful candidate will be expected to represent ISF with professionalism, positivity, and poise while describing technical concepts with simplicity and accuracy; the ability to communicate technical findings clearly to non-specialist audiences is vital to this role
- The successful candidate will be expected to be responsive, providing timely replies, proactively communicating blockers or schedule constraints, and comfort working within a government-contracted environment where deliverables carry external deadlines
- Access to project data requires current CITI Program certification in Human Subjects Research. Candidates without current certification must complete CITI training (self-paced, available at citiprogram.org) before data access is granted
- Doctorate preferred (Ph.D. or equivalent) in biostatistics, statistics, health data science, epidemiology, or a closely related quantitative field, combined with professional experience delivering quantitative and qualitative work in a client-facing or externally accountable context (e.g. a consulting firm, applied research organization, government advisory role)
- Experience working with or advising public sector or academic clients on quantitative and qualitative methodology, data systems, or analytical products, preferably in a public health or human services context
- AZ-204 certification is highly preferred
- Familiarity with version control (Git or equivalent) and the discipline of maintaining clean, reproducible, well-commented code. The ability to navigate and evaluate someone else's codebase is a meaningful part of the role. The ability to comply with the open science framework for reproducibility and traceability is preferred
- Proficiency in R, Python, or equivalent, including relevant packages for data manipulation, spatial analysis, and statistical modeling. Ability to work within RShiny and within an established Azure and Databricks environment following documented procedures is highly preferred
- Willingness to use Claude Code (Anthropic's AI coding assistant) as a productivity tool for reviewing scripts, running checks, and navigating the codebase. Prior experience with AI-assisted development tools is a plus
Benefits
- This is a 1099 independent contractor engagement. The contractor is responsible for their own taxes and benefits.
- Must currently hold, or have the ability to obtain, CITI certification to access restricted data
- Ability to meet internal deadlines as agreed with Principal.
- Available for occasional 1-hour meetings between 9 AM – 5 PM ET, particularly during onboarding
- Willingness to use Claude Code (Anthropic’s AI coding assistant) as a productivity tool for reviewing scripts, running checks, and navigating the codebase.
- Comfort operating as an independent contractor within a structured engagement: estimating and tracking your own hours, meeting deadlines with appropriate autonomy, and communicating proactively with a manager when scope or schedule questions arise.
- Access to project data requires current CITI Program certification in Human Subjects Research. Candidates without current certification must complete CITI training (self-paced, available at citiprogram.org) before data access is granted.
Company Overview