Building a company in Canada while driving for Uber
I landed in Toronto in August 2022 with a computing degree, a few years of Java work in Nepal, and a plan that was mostly "learn, then see". The learning went well: a postgraduate diploma in AI and machine learning, an internship, a lot of evenings reading papers. The "then see" turned into a decision I did not expect to make, which was to start a company instead of looking for a job.
I want to write about this plainly, because most founder stories you read are written after the outcome is known, and this one is not.
Why a company
Part of it was conviction about the problem. I had spent enough time building and evaluating AI systems to believe that the hard part of putting agents to work is not making them capable; it is making them accountable. That felt like a problem worth years, and the people I wanted to work on it with were not inside any company I could apply to.
Part of it was about where I live. Canada has been generous to me, and the honest way I know to return that is to build something here that employs people here. There is a lot of talk about talent leaving; I would rather be a small counter-example than a statistic.
And part of it, if I am being truthful, is temperament. I am thirty. I grew up in the eastern Terai of Nepal, in a region where most people my parents' age ran something: a shop, a farm, a school. The idea that your work is something you own rather than something you are given is not foreign to me. It just took a while to apply it to software.
What the days look like
Zerava Technologies Inc. is pre-seed. That is a polite way of saying the company does not yet pay me. So the week is split. Monday to Friday is the company: code, customer conversations, grant applications, the hundred small administrative things nobody tells you about incorporating. Weekends are driving — Uber and Uber Eats — which covers rent and keeps me honest about what work is actually worth. For a stretch I also did contract evaluation work for AI labs, which was useful in a way I did not expect: there is no better education in how models fail than being paid to make them fail.
I am not romantic about the driving. It is tiring and it is time I would rather spend on the product. But I have met more of this city from the driver's seat than I ever would have from a desk, and the conversations with passengers have taught me more about how ordinary people feel about AI than any survey.
What I have learned so far
- Focus is the whole game. We spent our first months with five product ideas that all shared a kernel. Each was defensible. Together they were a way of not choosing. The day we cut it to one — a governed agent runtime for organisations that cannot send their data to someone else's API — was the day the company started.
- Say the legal name. Small thing, but every form, every registry, every badge program wants "Zerava Technologies Inc.", not the shorthand on the website. Get it right once and you stop getting it wrong forever.
- Programs are worth the applications. Being accepted to Radical Ventures' AI Founders Masterclass this year, and joining the NVIDIA Inception program, did not change the product. They changed who would take a meeting.
- Learn from your co-founder, in public. Andrej wrote the research our platform is built on. I am the person turning it into something deployable, and I am still learning the mathematics from him. Pretending otherwise would be the fastest way to lose the trust of anyone who actually reads the papers.
If you are starting from the same place
If you are a recent immigrant with a technical background and an idea, here is the only advice I feel qualified to give: the gap between you and the founders you read about is smaller than it looks, and most of it is paperwork. Incorporate properly. Keep records from day one as if an auditor is coming, because one day someone will ask. Apply to every program you are remotely eligible for. And find a way to pay your rent that does not depend on the company succeeding this quarter, so that you can make decisions on their merits rather than out of fear.
I will write an update when there is something to report. Until then, I am in the car on Saturdays, if you want to talk.