A
Completed client project, publication approved
A finished engagement that the client has approved for publication, with the facts documented.
Our work
Research engineering work and demonstrations
This page explains how we label what we show, so you can tell a client project from an illustration.
Work is published here only when it is documented and, where it involves a client, approved. Numbers appear only when backed by documented evidence.
Labels
Three kinds of work, labeled exactly
B
Independent technical demonstration
A self-created example built with public or synthetic data. It is never a client result.
C
In progress
Shown only with explicit permission from the people involved.
Published work
Case studies
No client case study is published on this site yet. Each one will appear only after the client approves publication. When a case study is published, it will include:
- The problem and its context
- Permissions and the client's approval to publish
- Approach and architecture
- Validation method
- Verifiable outputs
- Limitations
- Technologies used
- Approved links
Independent technical demonstration
What an automated validation check looks like
A small demonstration of one kind of check used when financial data is extracted from documents.
| Entity | Assets | Liabilities | Equity | Result |
|---|---|---|---|---|
| Synthetic Co. A | 1250 | 780 | 470 | Pass |
| Synthetic Co. B | 980 | 610 | 370 | Pass |
| Synthetic Co. C | 1500 | 900 | 560 | Review: difference of +40 |
| Synthetic Co. D | 720 | 455 | 265 | Pass |
| Synthetic Co. E | 640 | 400 | 250 | Review: difference of -10 |
Rows that fail the check are queued for manual comparison with the source document. The corrected value, or the reason it stays unresolved, is recorded alongside the data. Checks like this are one part of validation. They do not make a dataset error-free.
Selected capabilities
What we are set up to do
These are capabilities, not a list of completed projects.
Research Data Engineering
- Document and PDF extraction into structured tables
- Cleaned dataset with data dictionary
- Versioned, re-runnable processing code
Research Analytics & Statistical Computing
- Analysis-ready datasets and scripts
- Tables and figures generated from code
- Model scripts with diagnostics
Research Software Engineering
- Analytical web applications and data apps
- APIs and system integrations
- Automated tests and technical documentation
Research Monitoring & Evidence Systems
- Validated indicator pipelines
- Survey and program data quality checks
- Dashboards and automated periodic reports
Have a project in mind?
Tell us about the data and the outcome you need.