Matt Thompson

The harsh reality about job searching as an experienced developer: Your years of experience can actually work against you. Here's why: • Companies assume you're "overqualified" for roles you'd love • Startups think you'll be too expensive or inflexible • Your resume gets filtered out because you're "senior" but the role says "mid-level" • Ageism is real in tech (even if no one talks about it) But here's what I've learned helping 100+ developers navigate this: The ones who succeed flip the script. Instead of hiding their experience, they leverage it strategically. They don't just say "I have 10 years of experience." They say: "I've seen codebases grow from prototype to IPO." "I've made every scaling mistake so your team doesn't have to." "I can mentor your junior developers while shipping features." The difference? Value over years. Impact over tenure. Results over resume bullets. Your experience isn't a liability. It's your competitive advantage. But only if you position it right. Here's how to reframe your experience: • Don't apologize for being senior • Show how you've evolved with technology • Demonstrate you can still be hands-on • Prove you're invested in continuous learning The market needs experienced developers who can: Think strategically AND code tactically. Mentor others AND deliver features. See the big picture AND sweat the details. That's you. Now own it. How are you positioning your experience as an asset, not a burden?

10

1 Comment

Coding agents make you faster. That part's obvious. The less obvious part is that speed without discipline just means you produce garbage quicker. I've watched teams adopt these tools and end up drowning in generated code nobody fully understood. The ones doing well aren't doing anything exotic; they review what comes out, they test it, they push back when the output is wrong. Basic stuff. Good engineering habits aren't optional just because a machine wrote the first draft. They're more important than ever.

17

Do architects inspect whether the lightbulbs work? Of course not. So why are engineers still the ones inspecting if form submissions work? Engineers didn't end up owning QA because they're the best people for it. They ended up owning it because nobody else could use the tools. Engineers are brilliant, but they're expensive, and they think in implementation details, not user experience. We hear the same thing constantly on sales calls: "We need a non-technical person to own and operate our QA process." That's not a niche request. That's the whole point. Prompt-driven testing changed that. You describe what you're trying to do in plain English and it runs. PMs can write tests. Business analysts can write tests. The people who think like your users can write tests. And that's how you actually find the weird stuff before it ruins your weekend.

5

1 Comment

I've been writing code in Cursor for a while. The weird part is I kept having moments where I thought "wait, what happened here?" You ask the agent for something and it jumps straight into code. Requirements were left half-done. Planning and implementation blurred together, and the next day you'd pick up the same project with a completely different standard. So I put together something small to fix that: Cursor Agent Kit. It's not an app. It's a `.cursor/` template you drop into any project. The idea is simple: clarify the prompt first, write a brief, make a plan, then code. Less "just build it," more "let's actually understand what we're building." What I like about it: - Vague prompts get clarified before anything else happens - You get a brief and a plan before code starts - No hunting for `@skill` every single time - Team defaults live in git, so sessions feel consistent - Works across Node, .NET, Python, Go — whatever you're on Setup is one command away: git clone https://lnkd.in/dz59Ap7S ./install.sh /path/to/your-project 💡 Tip: Don't want to install manually? In your project, tell the agent: "write the repo name, install it, and set up the monorepo structure too" — it'll be ready to go. I open-sourced it because I figured other people wrestling with the same Cursor workflow might find it useful. Link: https://lnkd.in/dMZaZZa8 Stars are nice. Comments are even better. Not on Cursor yet? If you want to try it with 50% off your first month, here's my referral link: https://lnkd.in/da6GaSzW

7

We’ve started wiring LLMs into production as if they were junior devs who can magically guess which of our 40 half-baked tools to call. Then we’re surprised when they do exactly what you’d expect with a messy, overloaded tool layer: random calls, wrong tools, brittle prompt dependence. In this new post I walk through a few recurring antipatterns in LLM tool design: - too many overlapping tools - the “god tool” that allegedly does everything - over‑nested, baroque schemas - vague, conflicting tool descriptions - hidden side effects that quietly corrupt state The second half is not about fine‑tuning the model, but about unit testing your tool layer: - treating tools as units with explicit contracts - writing prompt → expected tool call pairs - snapshot‑style tests on the tool invocation trace - and how frameworks like DeepEval can help here If you’re wiring LLMs into real systems, this layer largely decides whether the model looks “smart” or “unreliable”. Full article on Substack: 👉 https://lnkd.in/dGgQDC_P

3

Most coding bootcamps teach you how to build, but skip the reality of deployment. Your app’s sitting on a free subdomain, your users can’t log in, and you have no clue how much API calls are costing you. That’s a demo, not a product. This isn't theory. This is the Vibe Coding Blueprint – the stuff that makes your app *real*. We’re talking custom domains so your app has its own address, actual user authentication, cloud storage, chatbots, and a strategy to manage API credits so you don’t go broke. This is Week 3 energy, the practical skills that separate a project from a product. This is what it takes to ship. #AppDeployment #CustomDomain #BuildWithAI #NoCodeToProCode #APIStrategy #TechEducation #VibeCodingBlueprint

1