Research domains
Built for the hardest data in research
Six domains where evidence is difficult to collect, combine, and defend, and what we build for each.
The engineering is common across domains. The data, the definitions, and the risks are not, so each engagement starts from the domain.
01 / Domain
Economics & finance
Panels, filings, and administrative records that have to reconcile before anyone can model them.
The challenge
Economic research runs on data assembled from annual reports, regulatory filings, statistical releases, and administrative records. Layouts change, definitions shift, and a single misread unit can move a result. The work is as much reconciliation as analysis.
Related use cases
Data we typically work with
- Annual reports and regulatory filings
- Statistical releases and databases
- Administrative and transaction records
- Time series from multiple providers
What we build
- Entity-period panels built from document collections
- Accounting-identity and cross-year reconciliation checks
- Currency, unit, and scale normalization with versioned rules
- Reproducible data preparation for econometric work
02 / Domain
Public health
Surveys, facility records, and programme data linked across years without losing comparability.
The challenge
Health evidence is spread across household surveys, facility reporting, and programme systems, each with its own codes and reporting periods. Linking them responsibly means careful identifiers, privacy-aware handling, and clarity about what is truly comparable.
Related use cases
Data we typically work with
- Household and facility surveys
- Programme and service-delivery records
- Registry and reporting-system extracts
- Geographic and demographic reference data
What we build
- Harmonized multi-wave datasets with comparability registers
- Identifier crosswalks and join diagnostics
- Indicator pipelines with defined numerators and denominators
- Data-handling arrangements agreed before sensitive data is accepted
03 / Domain
Education
Assessment, enrolment, and survey data in many formats, combined into one defensible picture.
The challenge
Education research combines assessment results, enrolment records, school surveys, and administrative data. Formats differ by year and jurisdiction, and categories are renamed often. Without explicit mappings, trends can reflect a coding change rather than a real one.
Related use cases
Data we typically work with
- Assessment and examination data
- Enrolment and administrative records
- School and household surveys
- Policy and curriculum documents
What we build
- Cross-year, cross-jurisdiction harmonization
- Variable mapping with researcher-approved definitions
- Reproducible analysis workflows for recurring reports
- Document collections structured for coding and analysis
04 / Domain
Climate & environment
Large, messy time series from many sensors and sources, turned into datasets a model can trust.
The challenge
Environmental data arrives from stations, sensors, remote sensing, and reporting bodies at different resolutions and with different gaps. Quality control, gap handling, and metadata discipline decide whether downstream analysis is meaningful.
Related use cases
Data we typically work with
- Station and sensor time series
- Gridded and remote-sensing products
- Emissions and reporting inventories
- Regulatory and monitoring documents
What we build
- Ingestion pipelines with quality flags and gap reporting
- Unit, resolution, and calendar harmonization
- Versioned processing so results rerun exactly
- Analysis tools and dashboards for research teams
05 / Domain
International development
Partner-submitted data and indicators across many programmes, traceable from report to record.
The challenge
Programmes report through many partners, each with its own spreadsheets and definitions. Indicators are rebuilt every cycle, and funders ask where numbers came from. The need is a system where submissions are validated and every indicator traces to its source.
Related use cases
Data we typically work with
- Partner spreadsheets and templates
- Survey and monitoring exports
- Programme and beneficiary records
- Donor reporting requirements
What we build
- Validated submission pipelines with actionable exception reports
- Executable, versioned indicator definitions
- Deduplication and disaggregation logic
- Reporting outputs linked to calculation versions
Working in a domain not listed here?
The method transfers. Describe your data and the decision it supports, and we will tell you plainly whether we are the right fit.
06 / Domain
Social policy
Administrative data, surveys, and policy documents combined into evidence a decision can rest on.
The challenge
Policy questions draw on administrative records, repeated surveys, and large bodies of legislation and reports. The evidence has to be harmonized, documented, and reproducible, because it will be challenged by people with a stake in the answer.
Related use cases
Data we typically work with
What we build