One of the worlds top AI developers is leaving Google after close to a decade.

The 34-year-old French developer wrote in a post on X that he's starting a new company with "a friend," but that's about all he would reveal.

I'm thankful for my decade at Google, Chollet wrote in the post. "During that decade of time, deep learning moved from a niche topic in academics to being behemoth of an industry employing millions.".

Chollet is perhaps best known for creating Keras, a deep-learning-powered API created to empower high-level development of AI models and solve machine learning tasks. A blog post on Google's developer blog notes that Keras has more than 2 million users and powers many high-profile tech products-like Waymo's self-driving cars, as well as the recommendation engines on YouTube, Netflix, and Spotify.

In 2019 he published the Abstraction and Reasoning Corpus for Artificial General Intelligence (ARC-AGI) benchmark, which tests AI systems on novel reasoning problems. This year he launched the ARC Prize, a $1 million competition to beat ARC-AGI. (It remains unwon).

Chollet has often argued that the strategy pursued by many of the leading AI labs in the industry- throw ever more data and computational resources at models- will not get AI as "smart" as humans. In his opinion, probably the most promising approach is those methods that make models "reason" in a way that's closer to being human, such as neuro-symbolic AI.

In 2021, the Chollet was granted the Global Swiss AI Award for its significant research breakthroughs in AI. Time included him in its list of the 100 most influential people in AI in September.

According to Time, Chollet considers the kind of super intelligent AI as a tool for advancing human knowledge. "Artificial general intelligence is going to be a kind of super-competent scientist," he said.

Chollet said that Jeff Carpenter, machine learning engineer at Google, would take over as the new team lead for Keras.

"I have full confidence in Jeff and the incredibly talented Keras team to continue pushing the boundaries of what's possible in deep learning," Chollet wrote in his post. "I will stay deeply involved with the Keras project from the outside."