Staff Research Engineer

Onton•San Francisco, CA

80d•$195,000 - $230,000•Hybrid

About The Position

Our mission at Onton is to help people make decisions they love, instantly. We’re tackling the most economically impactful decisions first: the average shopping journey takes 79 days, and we’re taking that to less than 1. Since our public launch in late 2023, we've grown exponentially to guide millions of users to their perfect product. We recently built the foundation of a neurosymbolic AI agent that learns more about the world on every search. You have the chance to solve equally difficult problems as you scale Onton 10x over the next few months to serve 1B+ searches/year, which is just the beginning. This is only the beginning. Providing a person with sufficient context to make any decision in a truly optimal way amounts to creating AGI. The economic engine we’re creating on the e-commerce side, while a huge opportunity in its own right, is ultimately to support this end. Paradigm shifts like the one we’re in — from narrow AI to human-level AGI and beyond — are vanishingly rare. We want to offer you the chance to do your life’s work solving the problem that will solve all other problems. Together we will create an AI agent capable of providing a person, or itself, with sufficient context to make optimal decisions as quickly as possible. This agent will ultimately be capable of superseding most things we do at Onton. Our current application of e-commerce, making an optimal purchase decision — considering user preferences, millions of products, and billions of reviews — already requires a world model and a logical consistency far beyond the state of the art. Even OpenAI’s GPT-5.6 can’t rid itself of the problem of hallucination, and there are no sufficiently robust bolt-on solutions for context finiteness (RAG is not enough), catastrophic forgetting, or many other fundamental problems afflicting current-gen AI systems. Contra OpenAI, we believe that scale is not all you need. These challenges can only be addressed at the architectural level. Recently we’ve built the foundation of a neurosymbolic AI agent that learns more about the world on every search. It’s still in its infancy, but its compositionally-grounded architecture shows promise as the basis of a truly general problem solver. With your ingenuity — realized through your self-directed, data-driven experiments from first principles — we can turn its sparks of intelligence into a fire.

Requirements

Significant software, ML, or research engineering experience
Deep, principled understanding of the current state of the field of AI, especially neurosymbolic approaches.
Some experience contributing to empirical AI research projects
Enjoy implementation, not just writing papers.
Excited about non-purely-statistical approaches to AI.
Prefer fast-moving collaborative projects to extensive solo efforts.
Quickly and iteratively ship code that is correct, robust, scalable, and maintainable.
Full-time, in-person collaboration required for final interviews and initial role.
Ability to attend onsite interviews at the SF office.

Nice To Haves

PhD or Master’s in Artificial Intelligence, Computer Science, Cognitive Science, Linguistics, or comparable (or equivalent research experience).

Responsibilities

Create an AI agent capable of providing a person, or itself, with sufficient context to make optimal decisions as quickly as possible.
Contribute to empirical AI research projects.
Implement solutions, not just write papers.
Ship code that is correct, robust, scalable, and maintainable quickly and iteratively.