Earned, never bought
Our product's integrity is the business. We turn down revenue that would compromise the score — and we expect everyone here to defend that line without being asked.
Loading…
We're a small, distributed team making it possible for anyone — human or AI — to know which companies deserve their trust. If you care about clarity over cleverness and impact over optics, we'd like to meet you.
Every day, millions of people make decisions — where to buy, who to hire, which service to rely on — based on reviews they can't verify, ratings that can be bought, and search results shaped by whoever spent the most. At the same time, AI assistants have become the first stop for buying research, and they inherit every flaw of the data they read. The result is a trust crisis hiding in plain sight: the more the internet mediates our decisions, the less we can rely on what it tells us.
TrustRating is our answer. We combine verified human reviews with independent analysis from multiple AI models into a TrustScore that can't be purchased, gamed or negotiated. Businesses get a fair shot at earning a great reputation; consumers get an answer they can actually check; AI models get clean, verified evidence to reason from. It's infrastructure work — unglamorous, exacting and quietly enormous — and it's the kind of problem that rewards people who like being trusted with hard things.
Four principles that survive contact with real deadlines.
Our product's integrity is the business. We turn down revenue that would compromise the score — and we expect everyone here to defend that line without being asked.
The people who write the code also talk to the customers. Nobody ships a feature they wouldn't personally support at 2am — which is precisely why 2am rarely happens.
Simple systems, plain writing, honest numbers. If it takes a diagram, we draw the diagram. If it takes admitting we were wrong, we admit it fast.
Meetings are scarce and intentional. Deep work is the default state, written context is the norm, and time zones are a feature — the queue never sleeps, and neither does progress.
We are distributed on purpose. TrustRating was remote before it was fashionable and stayed remote after the pendulum swung back, because the way we work depends on it: long, uninterrupted stretches of focus, decisions made in writing where everyone can read them, and a team assembled for talent rather than postcode. Your day starts when your brain does. What matters is what ships, what gets fixed and what gets learned — not when your status dot turns green.
The work itself is unusually varied for a small company. In a single week an engineer here might tune the fraud heuristics that keep review bombing out of a small bakery's score, rebuild an admin surface that our own moderators live in eight hours a day, and design the API that a Fortune 500 uses to pipe verified reviews into its data warehouse. Because the platform spans consumers, businesses, moderators and AI pipelines, there is always a frontier that matches what you want to get better at.
We hire for judgment, not pedigree. Some of the best people here have no degree; others have PhDs. What they share is the habit of understanding a problem before solving it, the humility to write down what they don't know, and the spine to push back when something is wrong — including on the founders. Titles are thin, ownership is thick, and the phrase 'not my job' has no translation in our vocabulary.
Feedback here is a gift economy. Code review is thorough and kind. Post-mortems are blameless and specific. Praise is public and precise — 'good job' is banned in favour of saying what was actually good. Twice a year we write each other honest, structured feedback that goes to the person, not to a file. People grow fast here because the information they need to grow is never withheld from them.
The impact is unusually legible. Every improvement to review verification protects thousands of small businesses from sabotage they could never fight alone. Every millisecond shaved off the public pages is multiplied across millions of visits from people making real decisions with real money. Every fairness fix in the score touches livelihoods. It is very hard to work here for a month and wonder whether your work matters — the dashboard answers before you finish asking.
And we look after people properly: meaningful equity, hardware you choose, a learning budget that renews without ceremony, and genuinely flexible time off that leadership visibly takes too. We are not a family — families don't do performance reviews — but we are a team that keeps its promises to each other, and that turns out to be rarer and more valuable.
Work from anywhere with 4+ hours of overlap with Central European mornings.
Everyone owns a piece of what they're building. Standard, founder-friendly terms — for you.
Pick your hardware and tools. We buy them without a procurement committee.
Books, courses, conferences — approved by default, renewed every year.
Rest is part of the job. Leadership takes real holidays and expects you to as well.
A calendar you control. Meetings are the exception and always have an agenda.
Four steps, two weeks on average, and you'll always know where you stand. We pay for take-home work because your time is worth money.
Thirty minutes, two-way. You interview us as much as we interview you — bring hard questions.
A paid take-home or a live pairing session — your pick. Real work, no puzzles, no whiteboard theatre.
Meet three or four people you'd actually work with. We're assessing fit both ways — so are you.
Usually within a week of the team day, with full context on compensation, equity and what your first 90 days look like.
We hire engineers, ML researchers, customer-trust specialists and designers — usually before the job post exists.
No listed openings right now — but the best people we've hired wrote to us anyway. If you're exceptional at what you do and the mission resonates, don't wait for a job title to appear.
Tell us what you do best and what you'd want to build here. A short, honest email beats a long, polished one.
Trust is the internet's oldest unsolved problem. Come solve it with us.