Alen Sevšek
I bridge science, product, and go-to-market, connecting dots most careers keep apart.
A decade translating science into adoption taught me that trust is the actual product. This site is the proof beside my CV: products I actually built and a business I co-run on AI, each designed so the honest version is the shipped version.
I was trained as a medicinal chemist, which is a long education in one question: is this claim actually true? A decade of commercial work followed, taking biosensing, contract research, and IoT software into markets where a wrong claim costs more than a lost deal. I now build my own products with AI, and the question has not changed.
Three people who watched this work up close put it on record in writing: a PhD supervisor, a CEO who has hired scientists for twenty years, and the managing director of a software firm. The strongest lines are below; each links to the full letter.
V Á K U U M
LoreTrip
SupplementPeer
Mijn Slovenië
V Á K U U M
LoreTrip
SupplementPeer
Mijn Slovenië
I am a domain expert who ships software, not a career developer. I design the product, set the architecture, and direct Claude Code and Claude Design as the execution layer: they write most of the code, I read it, question it, and decide what ships. The judgment about what to build, what is safe to rely on, and what is honest to claim stays with me.
The method is visible in the work: a Postgres two-schema design with build-enforced separation between rating and commerce logic, CI verification tests that fail the build when that boundary is crossed, a headless-Chrome PDF pipeline, MCP connectors, and a multi-tenant white-label architecture that re-skins one codebase per client. I could not hand-write all of it. I can read all of it, and I can tell you why every piece is there.
Doctoral Research
A new class of guanidinium iminosugars built to inhibit glycosidases with a selectivity natural-product inhibitors rarely achieve. Three first-author papers. Selective versus merely potent is the thread through everything I have built since.
Glycosidases sit at the center of how cells process sugars: when they fail, disease follows. My doctoral work designed and synthesized a new class of guanidinium-bearing iminosugars, chemical tools built to inhibit these enzymes with a selectivity that natural-product inhibitors rarely achieve.
Career Readout
The record below is chronological. As each era enters view it deposits its skills into the ledger, reading by reading; by the last era the whole stack sits together.
Twenty years on one arc: science → product → adoption. Five years of doctoral medicinal chemistry, then a decade taking science to market: SPR biosensing across four continents, product ownership, a CRO built from zero, and the first commercial hire at a 40-person software company. Now building applied-AI products end to end. Three executives have put this record on record.
{{ era.summary }}
Good AI deployment is judged by who is better off after it ships
Drug development has a standard that sounds obvious and is hard to meet: a potent compound that cannot be dosed safely is not a product. Potency is the headline number, the thing that looks impressive on a slide. Dosing is where the compound meets a liver, a kidney, a patient who also takes three other medications. The field learned, through decades of expensive failures, to celebrate potency quietly and reserve judgment for the dose.
Applied AI is at the potency stage of its story. The capabilities are real and improving, and most of the attention sits exactly there, on what the model can do. I hold my own work to the other standard: good deployment is judged by who is better off after it ships.
Journeys
