Quantitative Researcher

LaurenceZürich, ZurichOn-siteFull-timeNew grad, 0–1 yearsListed 9 hours ago

Apply now

About this role

Laurence (YC W26) is the frontier operating system for marketers. We run brands’ advertising with quantitative models and AI, with profit at the center. Brands with $10M to $100M in annual revenue, including Pit Viper, Fi, and SPYLT, trust Laurence with their advertising, and customers typically increase profit by 15% to 40%.

Under the product is a research problem: what is the next click, the next creative, or the next dollar of budget actually worth to this business? We answer it with models that treat every estimate as a probability distribution, so risk is priced into every decision, every hour.

We’re building our research team in Zurich.

The team

Our CEO, Matthew, was a machine learning engineer at Meta and Google. Our co-founder, Leo, was a quant at Jump Trading and a high-stakes poker professional. Our early team includes IMO and IOI medalists and people from Jane Street, Five Rings, BCG, and Salesforce.

Everyone here works on the hardest problem in front of them, and you’ll learn faster here than anywhere else you could go.

In Zurich, you’ll work directly with Leo and our research team on the models that move our customers’ money.

The role

Your models decide where real money goes. In practice, that means:

* **Own the models that set every bid.** Conversion-rate estimation, bidding policies, budget allocation, and the signals that feed them, running live across our customers’ Amazon accounts.
* **Take the models to new channels.** Extend what works on Amazon to Meta, where creative and audiences matter as much as bids.
* **Model consumers before they act.** Help build Laurence Simulations, which predicts how customers will react to products, creative, and copy before anything goes live.
* **Test on live auctions.** Design experiments where real spend is the readout, and simulations that test a policy before it touches customer money.
* **Ship your own research.** Research here ends in production, not in a notebook: the system reprices every keyword hourly, and your models drive it.

What you bring

* Depth in probability and statistics; Bayesian modeling is a working tool for you, not just a course you took.
* You can tell a real effect from noise, and you design experiments that settle the question.
* You write production-quality Python and are comfortable owning a model from data to deployment.
* Taste for the simplest model that is correct, and honesty about uncertainty.
* High agency and learning speed; comfortable with direct critique and an intense pace.

Experience with auctions, advertising, reinforcement learning, or LLM-based simulation is useful but not required. A PhD is welcome; what matters is research you have taken all the way to a decision.

How we work

We care more about the truth than who is right. If you think a founder’s assumption is wrong, say so and show your reasoning. Expect the same openness when someone challenges yours.

We move quickly and take responsibility seriously. Grit shows up in the follow-up, the second investigation, and the difficult conversation you don’t put off. Agency means taking a useful first step and asking for help before you get stuck.

We’re honest with each other, own our mistakes, and help each other improve. We care about the people relying on our work.

Our values

* **Relentless truth-seeking.** We are approaching this problem in a way nobody has before: there is no playbook, and there is no beaten path. We need to iterate toward the truth, both actively accepting uncomfortable realities and readily rejecting simple fictions.
* **Strong beliefs weakly held.** Be intensely convicted while also being eager to be convinced. Innovation is earned: 90% of our ideas will turn out to be wrong and few will turn out to be right, but we need to execute with 100% dedication to tell which is which.
* **Devotion to one another.** People fall; pick each other up. At the same time, be the best version of yourself: not just in terms of your individual actions, but how you exist and interact with a world of other similarly driven people.
* **Default to “yes”.** If in doubt, believe your teammate until proven wrong. At worst, you’ll end up being right and your teammate learned something new. At best, you’ll have learned and we will all be better for it.
* **Commitment to growth.** At every point in time, we exist as the worst versions of ourselves compared to all future selves. This is a blessing: try something new, take a risk, push yourself and take a shot in the dark. Inaction is the thief of betterment: treat every day as an opportunity to grow, and seize all the opportunities you get.