Chip Huyen
9 Intel
Chip Huyen is a Vietnamese writer and machine learning engineer who grew up in a rice-farming village in Vietnam and studied at Stanford, where she created and taught courses on TensorFlow for deep learning and ML systems design. She has worked at NVIDIA, where she was a core developer on the NeMo framework, as well as at Snorkel AI and Netflix. She is the author of the bestselling books "Designing Machine Learning Systems" and "AI Engineering," alongside several earlier Vietnamese story collections.
Is this you? Sign in with X to claim this profile.
Intel
Ten major open research directions in language models, drawn from conversations across industry and academia, with hallucination and context learning the most discussed and multimodality and new architectures the most promising.
Chip Huyen covers AI engineering fundamentals, why many companies struggle to ship AI products, how hard coding-agent productivity is to measure, and how to improve apps via user feedback.
How generation is actually configured: temperature, top-k and top-p as ways of shaping a probability distribution rather than magic dials, and how test-time compute fits alongside them.
A primer on multimodality and large multimodal models: why systems that only read text are a poor fit for a world humans navigate by reading, listening and seeing at once, and how such models are built.
What a survey of the 900 most popular open-source AI repositories shows about how the ecosystem has changed, four years after the same exercise on the open ML landscape.
Instead of ranking models and picking one, predict which model a user would prefer for a specific prompt, turning human preference data from a leaderboard into a routing signal.
The mistakes that recur when teams build with foundation models, starting with the most common one: reaching for generative AI where it was never the right tool.
The components that recur across generative AI platforms once you look at how companies actually deploy them, built up from the simplest possible architecture rather than presented as a finished diagram.