Amazon's HR chief wrote 100,000 lines of code with AI after 25 years away from coding

[James90]

Beth Galetti, Amazon's senior vice president of People Experience and Technology, wrote in a company blog post this week about rediscovering software development through AI tools after not having written a single line of code herself in 25 years. Galetti started her career as a software developer building package tracking systems before eventually moving into senior leadership roles, and she describes feeling truly nervous starting her first project back, a simple family calendaring app, admitting it had been so long since she'd built anything that she wasn't entirely sure what to actually expect from herself going back into it.

That first small project apparently gave her enough confidence to go considerably further. On a single flight back to Seattle, Galetti opened Kiro, Amazon's own AI powered development tool, and by the time the plane actually landed she'd built the complete submission app for an internal team competition called Everyone Can Build, which challenges employees to build something that solves a real problem using AI regardless of their own individual technical background. The competition ended up drawing more than 1,500 participants within just a few weeks of actually opening it up, with one employee building an app that turned a nearly four hour manual process into a one second automated check, saving tens of thousands of hours annually across the team.

Galetti says she's since gone on to write more than 100,000 lines of code total using AI tools, and frames the broader shift as these tools really expanding what employees are actually capable of doing themselves, rather than simply replacing the specialized technical work that used to require a dedicated engineering team and an actual spot on that team's roadmap before anything could get built at all. Amazon's internal AI toolkit includes Quick, an assistant for data questions and research, Kiro for actually building things from a plain language description, and Aza, an internal workplace assistant that helps employees navigate things like benefits, IT questions, and internal job postings.

The blog post also details Amazon's broader 2.5 billion dollar Future Ready 2030 commitment, aimed at helping at least 50 million people worldwide prepare more directly for the future of work through education and skills training. That commitment includes a separate 1 billion dollar investment specifically in Career Choice, Amazon's existing employee education benefit, with a stated goal of upskilling another 500,000 employees globally in areas like cybersecurity, software development, logistics, and renewable energy, alongside making one AWS AI certification exam completely free for every single Amazon employee each year

I'm an Amazon SVP who hadn't coded in 25 years. Then I wrote 100,000 lines of code with AI.www.aboutamazon.com

[Niamh_25]

1,500 employees actually participating in an internal building competition within just a few weeks is a actually striking adoption number for something this new and unfamiliar to most non technical staff. Says a lot about how much real, latent demand for actually building things exists across a large company once the technical barrier to entry genuinely gets lowered enough

[Warden]

The four hour manual process turned into a one second automated check example is clearly the single most concrete, compelling illustration buried in this whole piece. That's properly real, measurable time and money saved directly, not just some vague, abstract corporate talking point about efficiency gains

[InferenceLoop65]

Worth being at least a little skeptical about how representative this particular story actually is of most employees' genuine real world experience with these exact same tools though. A senior executive with direct executive backing and genuine dedicated time carved out to experiment freely is a meaningfully different situation entirely than a regular frontline employee squeezing in the exact same kind of exploration around an already packed daily workload