Spark creativity with AI
Our vision is to help the world create with AI.
Building Hotpot demands a daunting number of tasks. Instead of mastering every detail, we want to partner with experts so we can focus on what matters to Hotpot users: product. Every minute spent duplicating someone else's skills is a minute not spent on developing features and enhancing the Hotpot experience.
Culture is everything.
One key difference between applied and academic labs arises when forced to trade-off between impact and novelty: one prioritizes impact and another novelty. This tension is illustrated by a wonderful anecdote by Ravid Shwartz Ziv who chronicles the roadblocks in reaching 1000 citations for a highly influential paper.
We focus on the former as a computer vision-language lab. Specifically, unglamourous and non-novel work may trump exciting architectural experiments because they benefit users more, even though they will not yield conference awards. Evals and data curation are two great examples. The work is mind-numbing and tiresome, and does not require a Ph.D., but it is required to do research here.
We are firm believers of simple ideas grounded in theory. Simple ideas may not win Turing awards, but they may win 1 billion users.
People like technology providers like doctors: non-partisan. At Hotpot, we are technologists, not activists. Our customers come to Hotpot for AI services, not to hear political beliefs or societal grievances.
The world faces countless challenges, from climate change to rampant disease.
It would be incredible if we could solve them all, but we are not talented enough. Moreover, lack of focus is the cardinal sin with startups.
The only way to maximize our odds of success is with maniacal commitment to a narrow slice of problems and with the sobering humility that endeavoring to please all only pleases none.
If we're extraordinarily focused and fortunate, we will make users happy and earn their business.
Accordingly, we are politically neutral and dedicated to solving issues at the technology layer, not the human layer. The printing press, the smartphone, and other general technologies both harm and help society. Imagine if governments limited smartphones as a misguided form of protection.
History and logic unequivocally prove that technology is the most powerful force for elevating living standards and advancing society. Our mission is to play a tiny role in advancing this trend.
We prefer people with minimal experience but provable learning capacity.
In fast-moving water like ML research, too much experience is more like an anchor than a propeller because many "facts" are merely weak assumptions masquerading as truth. Pre-GPT notions on scale are arguably the most prominent illustration of this.
More critical than experience are ability to learn and skills in foundational domains like math and computer science.
Because Hotpot is restricted from sharing GenAI research, we give back to science via an informal open research group called HotpotBio. The mission for HotpotBio is to collaborate with healthcare professionals from leading institutions like Stanford and release research in pursuit of AI doctors and curing cancer. See the HotpotBio repo for details.
Email info at hotpot.ai if you're excited about our mission and augmenting 1 billion people with AI. Explain which Hotpot products interest you most, and include the position-specific information below.
ML researchers: share the 5 most recent papers you read and one common assumption you think may be wrong (e.g., diffusion models should use Gaussian noise). No assumptions to challenge? No worries. This is the reason to join us -- to discover new truths and expose flawed assumptions. Hotpot provides a unique opportunity to both pursue SOTA research and directly benefit people. Impactful ideas won't sit up locked up on arXiv. Instead, they will flow into our AI products and benefit users around the world.
Developers: share resume and profiles from GitHub, Stack Overflow, and Hacker News.
Designers: share portfolio and resume. We're undergoing a massive facelift. Now's the time to join and leave your design mark on Hotpot. Please send proposals on how to make Hotpot look amazing.
Writers: share one popular article you thought was beautifully written but logically flawed.
