In the world of technology, there are people who write code and people who change the way computers think. Jeff Dean belongs, without a doubt, to the second group.
Present at Google since 1999 (he was employee number 30),
Dean is the legendary figure who has designed the infrastructure that today allows artificial intelligence to be an everyday reality for billions of people.
Long before we talked about ChatGPT or Gemini, Dean was already optimizing Google’s search and advertising systems.
However, his true legend in the industry began when he co-founded Google Brain, the research team that rescued deep learning from academic oblivion to the dominant force it is today.
The man behind TensorFlow and TPUs
If you’re wondering why AI has advanced so quickly in the last decade, the short answer is that Jeff Dean built the right tools.
He was the main driver of TensorFlow, the open source library that standardized the creation of AI models across the planet.
But he didn’t stop at software; aware that traditional processors were falling short,
led the design of TPUs (Tensor Processing Units), chips created exclusively to accelerate AI.
It is impossible to understand
Who’s who in AI?
without stopping at its ability to scale systems. Dean is the one who made it possible for massive models to be trained on thousands of machines simultaneously without failing.
In fact,
It was he who proposed the name Gemini for Google’s most advanced model
, symbolizing the joining of forces between the Brain and DeepMind teams under his supervision as Chief Scientist.
The legend and the coffee
Beyond its thousands of patents,
Jeff is known for his simplicity and his love of coffee
espresso (he is said to have made more than 20,000 cappuccinos for his colleagues).
He is the type of leader who still reviews lines of code personally, always seeking elegance and efficiency.
His trajectory reminds us that behind every intelligent response of a system, there is a massive engineering scaffolding and someone with enough vision to have built it from scratch.
This post is also available in: