Want to upskill in AI? This is what a former Nvidia engineer recommends (www.businessinsider.com)

Chip Huyen, a former Nvidia engineer who worked on the NeMo platform and taught ML at Stanford, says the fastest way to upskill for AI is to build end-to-end: pick an idea, implement it, and deploy it so someone can actually use it. She emphasizes practical exercises (e.g., a week-long log of everyday frustrations to pick a solvable project) and notes that AI coding agents now let even non-coders prototype useful tools — which rapidly builds confidence and practical understanding. Huyen also urges learners to balance hands-on work with structured study of foundations and tools (courses, books, curated curricula), a theme she expands on in her book, AI Engineering. For computer science students and engineers, Huyen stresses systems thinking over isolated coding skills: CS is about designing solutions across components and trade-offs, not just writing snippets. Technically, that means mastering deployment, integration, monitoring and being able to reason about AI-generated code — a skill Stanford’s Mehran Sahami and Andrew Ng also warn is critical. The implication for the workforce is clear: generative AI will automate many coding tasks, increasing demand for engineers who can architect whole systems, review and validate AI-produced code, and combine disparate knowledge to solve larger, more complex problems.