Analysing data without giving people away. I advise on anonymisation and set up Open Diffix for it, free software that allows queries while keeping the individuals inside them unrecognisable.
What it is for
For public bodies, companies and organisations that process personal data and still want to learn something from it: publishing statistics, passing on analyses, working with third parties.
The distinction everything hangs on
Anonymised is not the same as pseudonymised, and the difference matters legally. Pseudonyms can be traced back, true anonymisation cannot. Which one you need depends on what is meant to happen with the data.
How I go about it
First the stocktaking: which data, which purpose, which rules, and what risk you carry today.
Then the strategy. How much protection is needed, how much meaning should survive, and with which method that trade-off becomes something you steer rather than guess.
Then the setup. Installing Open Diffix, adapting it to your data and hooking it into the analyses you already run.
Then the handover. An introduction for the people who will use it, and working sessions on real datasets rather than examples.
Finally the check. Testing whether the result holds up, and adjusting when your data or the legal situation changes.
What is not part of it
No legal advice. I tell you what the method achieves technically and where its limits are. Whether that is enough for your case is decided by your data protection officer or your lawyers.
No guarantee of zero risk either. Anonymisation is a balance, not a switch. Anyone promising you absolute safety has misunderstood the question.