Comment by RomanPushkin

Funny story: in 2025 I interviewed with multiple Ruby/Rails companies in San Francisco. I performed really well in the interviews. For one company, I went through a full on-site interview at their office. And still — no offer. The interview went great.

I have ~20 years of experience. I wrote a book about Ruby. I have GitHub repositories with thousands of stars. I have my own successful projects written in Ruby/Rails. I’ve spoken at conferences and contributed a lot to the Ruby/Rails community. I was a perfect match — and I still wasn’t hired.

This wasn’t a one-off. The same thing happened with several Ruby/Rails startups.

You know what I did next? I switched to Gen/Applied AI. And the difference was huge. The feedback became much better, and salaries were 25–50% higher. The tech itself wasn’t that different — mostly dynamic languages. I had to learn new things, but it took months, not years.

I also pushed myself deeper toward understanding AI properly. I genuinely enjoy this space. I started learning the fundamentals and even built my own learning materials (for example, howllmworks.com). You don’t need to go that deep to get hired, but I wanted to. The field is fascinating.

What’s funny is that many companies hiring “AI engineers” don’t really know what they’re doing. I’ve had interviews where they openly said: “We don’t really have AI expertise, but we know we need AI.” That’s how things are right now. It’s both good and bad. They can’t really judge your skills properly — but that also means your chances of passing are higher.

As for the Ruby/Rails world — I’m honestly very disappointed. The market feels completely saturated. There are too many experienced engineers competing for too few roles. Being good is no longer enough.

One company literally told me my interview performance was too good. They suspected I was using AI. That was the feedback. Twenty years of experience, open-source work, a published book — none of that mattered. “You’re too good, and there are too many candidates like you.” That’s how I understood it.

I’ve seen this happen repeatedly. It’s not just one bad experience.

At this point, I genuinely believe the Ruby/Rails ecosystem is shrinking. The whole “one-person framework” idea that DHH has been promoting made sense years ago, but not anymore. The problems it was solving simply don’t exist in the same way today.

With LLMs, the world changed. You take the best tools available. Next.js with standardized React components instead of Stimulus and Turbo. Hosted auth instead of rolling your own. When I needed to integrate something like Clerk, I just dropped in a component and moved on. There are tons of ready-made solutions in the React ecosystem.

Now compare that to Ruby. Are there modern AI libraries? Yes, technically. Are they well-maintained? Not really. You’re often dealing with abandonware. LangChain officially supports Python and TypeScript — not Ruby. And like it or not, AI today is happening in Python.

The more time you spend clinging to Ruby/Rails, the further behind you get. That’s just reality. My advice is simple: if you can, move on. The opportunity window in AI is wide open right now, but it won’t stay that way forever. 2026 is probably the last really good entry point.