Medical school leaves little time. So instead of doing the repetitive work, I build systems that do it — and keep running when I can't.
Simulated log — illustrates the real collector/scoring/delivery flow, not a live feed.
An automated intelligence system that finds and ranks opportunities matched to a profile — freelance work, jobs, scholarships, certifications, conferences, and calls for projects.
Finding the right opportunity means checking a dozen sources every day, and most of what you read doesn't apply to you. That's hours a week spent filtering, not applying.
Sentinelle collects continuously, classifies what each item actually is before judging it, scores it against a live profile, and pushes only the top of the pile to a Telegram bot — with the first concrete action to take.
Free AI tiers run out. So scoring runs as a cascade: if the primary model is unavailable, it falls through to a backup, then to a keyword layer that scores below the insertion threshold on purpose — a degraded cycle inserts nothing rather than inserting noise. The system never silently lowers its own standards.
Every piece here feeds another. That's the whole operating principle — nothing is built to stand alone.
Client work pays for the tools and the time everything else depends on.
Sentinelle finds the work so the hours go into doing it, not searching for it.
Clinical training is what makes the eventual target — digital health — credible rather than aspirational.
Medicine trained me to look for the mechanism, not just the symptom. Sentinelle is what happens when that habit meets code — and design is what happens when it meets a blank page.
Hope is not a strategy.
Happy to talk through automation, scraping, or getting a first system running.
💭 Not ready yet? Just chat about an idea