Learn Chemistry with Cartesian Tutor

Cartesian Tutor is built by Brian Lee, original author of the ChemWOOT class (2013), three-time USNCO study camp mentor, MIT/Harvard graduate, and silver medalist at the 2007 International Chemistry Olympiad in Moscow, Russia. Brian has extensive experience in tutoring and mentoring elite students, through his direct work with top students at the camp (and continuing mentorship through college and post-graduation).

Cartesian Tutor is an experiment: we know that tutoring is by far the best educational methodology, but completely impractical at scale. Can we scale Brian's 1-on-1 tutoring reach by injecting his chemistry ideas and insights into an LLM? Brian's personal bar for success is an AI tutoring experience that is at least 75% as good as being directly mentored by Brian, at an accessible price point.

Cartesian Tutor's methodology is laid out in Brian's guide to studying for the USNCO. In short: take practice exams under realistic conditions to discover areas of weakness, do deep-dives to thoroughly understand the theoretical underpinnings, and then drill problems to build muscle memory.

Cartesian Tutor's features are built around this methodology:

Behind this methodology is a simple observation: every student comes in with a different background and familiarity with topics in chemistry (and more broadly, familiarity with related topics in science and math). A traditional classroom lecture setting forces advanced students to walk, not run, at the pace of the slower students, and wastes additional lecture time ensuring that the course is a self-contained, self-bootstrapping unit that teaches its own prerequisites. Instead, Cartesian Tutor starts at the finish line and works backwords, giving everyone a customized experience. This is accomplished by relying on a mix of AI's infinite patience to fill in each students' missing "background" material, plus Brian's lecture notes for teaching Olympiad-level Chemistry.

Not all features are fully built out yet - human-verified solutions to all problems are currently proving to be a pretty labor-intensive undertaking. Cartesian Tutor also does not anticipate offering any Organic Chemistry tutoring in the near future, as LLM performance is currently far below what is necessary to teach it properly.

About Brian

Brian (Bergen County Academies '07, MIT '11, Harvard '13) competed in the USNCO and made it to the study camp in 2006, and made it to the US team in 2007, winning a silver medal at the IChO in Moscow. Afterwards, he majored in chemistry, doing undergraduate research in organometallic catalysis, and then started a PhD studying biophysical models for the origin of life at Harvard University. After dropping out of his PhD studies, he went on to work in software, machine learning, and AI at Google Brain, Databricks, and other top tech companies. At Google, Brian worked on graph neural networks for scent prediction and advised on the feasibility of various climate initiatives (as they pertained to chemistry and/or ML).

Brian runs a personal blog at Modern Descartes, and has written extensively on chemistry topics. He has a very cute and precocious kid and refuses to build anything that he wouldn't want his kid using.