About

Sebastian Florescu, PhD

Medical writer and science communicator. Author of Everyday AI.

Portrait of Sebastian Florescu

I’m Sebastian Florescu, PhD, a medical writer and science communicator. I trained as a scientist and did my doctorate in cancer biology at the German Cancer Research Center. Today I work in Pharma and write the plain-language documents people read before and after taking part in a clinical trial. The consent form explains the key parts of a trial before someone decides whether to join: why it is being run, how it is designed, which treatments it involves, and what taking part asks of the people in it. The summary of results comes at the end and explains what the study found, from how well the treatment worked to how safe it proved to be. Both are complex documents, and both have to work for someone with no medical training. Written well, a consent form is the difference between an informed decision and a signature.

My work also reaches upstream, into Digital Data Flow: the industry’s move to define a clinical trial’s protocol, its master plan, as data rather than write it as a document. I help develop an end-to-end concept for it, in which the protocol, the consent form and the registry entries all flow from one study definition instead of being retyped for every trial. I also work on automations and AI-enabled tools that support the shift. AI has a real job in that picture, and so does the person who checks its work.

I’m also the author of Everyday AI: A Simple Toolkit for Real Life, a calm, no-hype guide for busy people who want AI to make everyday life lighter, not louder. It runs on few tools, clear jobs and a human in charge of the result, the same rules I follow whether I’m drafting with AI or building my own agents. Different audiences, same job: take something complicated and make it usable.

Expertise

What I work on

Plain-language medical writing.

Consent forms, trial summaries and lay summaries of results: the documents trial participants actually read. Short words and short sentences help, but only so far. The meaning has to stay right, down to how certain a finding is, and the reader has to understand the information well enough to use it. So I write from the questions a participant brings. Readable is the starting point. Usable is the goal.

Clinical documents as data.

What a participant is told should trace back to the study data it comes from, instead of being retyped for every trial. I help develop an end-to-end concept for how that could work: one structured study definition from which the protocol, the consent form and the registry entries would be generated, built on ICH M11 (what a protocol contains), CDISC USDM (the data model beneath it) and HL7 FHIR (how it moves between systems).

Generative AI in practice.

AI-assisted drafting at production scale, with a person reviewing every output that matters. I build grounded agents: compiling the sources, writing the instructions, and drawing a hard line between what the agent may claim from a source and what it may work out on its own. Just as important is knowing where a model belongs in a workflow, and where it has no business at all.

Visual communication.

Infographics and medical illustration, designed in Illustrator, Photoshop and InDesign, plus short patient-facing videos. In a document, a graphic rarely replaces the words. It backs them up: it anchors a key idea, helps readers find their way through a long text and makes the hard parts easier to understand. Some ideas still land faster as a picture than as a paragraph.

Background

From the bench to the page.

I studied biochemistry at the University of Bucharest, with an Erasmus year at the University of Regensburg, then took a master’s in molecular and cellular biology at Heidelberg University and a doctorate in cancer biology at the German Cancer Research Center. Then I moved into medical writing, and took the science with me.

Romanian is my first language. I work in English and German.

Selected work