harmonicr
From source systems
to sound decisions.
harmonicr builds governed data platforms, runs careful statistical analysis, and reports findings in plain language. Everything we deliver is reproducible, documented, and yours to keep.
From source systems to sound decisions.
Most engagements follow the arc below — from raw feeds through a governed platform to analysis and reporting. Scroll through it; the panel alongside shows one dataset moving through each stage.
Connect the sources.
Data rarely lives in one place. We connect the feeds — REST APIs, database extracts, laboratory and instrument outputs, scheduled file drops — and document each interface as we go, including the quirks.
Land it, unaltered, in a governed lake.
Source data is preserved exactly as received in an auditable landing zone. Nothing is overwritten, and every record can be traced back to the feed and date it arrived — the foundation of defensible reporting.
Refine through deliberate layers.
We structure the platform along medallion lines: bronze preserves the raw record, silver validates and conforms it, gold aggregates it for purpose. Each layer is defined in code — versioned, reviewed, and re-runnable.
Publish report-ready views.
Analysts and reporting tools consume consistent, documented views rather than ad-hoc extracts. Definitions are agreed once, encoded once, and reused everywhere — so two reports never disagree about the same measure.
Analyse with methods that fit the question.
Health and biological data bring seasonality, censoring, and structure that generic approaches miss. We select methods to match — and report uncertainty honestly, including when the data cannot support a firm answer.
Report for the people deciding.
Findings arrive as a clear document in plain language, alongside the complete code that produced every figure. When a stakeholder asks where a number came from, the answer is specific and checkable.
Capabilities across the full data lifecycle.
Each capability below reflects work we have delivered end to end, in R and on modern data platforms. If your requirement sits between categories, we would still like to hear about it.
A small practice with firm standards.
Three commitments apply to every engagement, regardless of size.
Show the working
Every figure traces back to code that can be run, read, and reviewed. If we cannot demonstrate how a number was produced, we do not report the number.
Advise candidly
If the evidence does not support the conclusion you were hoping for, we will say so — early, clearly, and with the analysis to back it up.
Build your capability
Documentation, code review, and mentoring are part of how we deliver, not an optional extra. Each engagement should reduce your dependence on the next one.
Recent projects.
A rotating selection — open any project to read the full solution paper.
The people who do the work.
harmonicr is deliberately small. The people below scope the work, deliver it, and remain your point of contact throughout.
Tell us about your data challenge.
Describe the situation in your own words — no preparation required. We will reply with an honest assessment, including whether we are the right fit for the work.