AI Engineering · 5 min read
What AI Coding Agents Change About Software Engineering
Hi, my name is
I'm a software engineer from Nepal 🇳🇵, now based in Japan 🇯🇵, with over nine years of experience building web applications, APIs and backend systems.
AI coding agents now write a growing share of the code. I believe the craft is shifting from typing code to specifying, reviewing and operating it, which makes fundamentals, tests and judgment more valuable, not less.
This site is where I write about that shift, and where I host Tech Weekly, a weekly podcast on software engineering news.
When an agent writes the code, you still have to know whether it is right. Data structures, complexity and system design are how you judge it.
Generated code is a draft from a fast, confident junior. Review it for correctness, security and fit, and own what ships.
Clear requirements and executable tests are how you hand work to an agent and know when it is actually done.
LLM features need timeouts, retries, queues, cost limits and observability like any other dependency, only more so.
A short weekly roundup of software engineering, DevOps and AI news, with the context on why it matters to people who build software.
Episode #47 · Feb 23, 2026 · 7:56
Episode #46 · Feb 15, 2026 · 7:27
Episode #45 · Jan 25, 2026 · 6:51
August 2024 - Present
06. What's next?
Want to talk engineering, AI tooling or the podcast? My code is on GitHub, and I'm easiest to reach on LinkedIn.