Stan Kolotinskiy

I was working for years in a huge product with a lot of legacy functionality, and I can tell that pretty much no one was able to make decisions that were based on a full understanding of the product, because, well, no one did :) The product was split into several relatively independent parts, so this approach worked quite well and didn't hurt too much

I can't unveil the actual invoice amounts, but every major cloud provider is providing recommendations about more efficient resources usage, so I'd strongly advise keeping an eye on it - that's what we're doing as well

No way I'd delegate that to an AI builder. I'm using LLMs extensively (primarily Claude with Claude Code) as a daily driver, but I wouldn't let it build something for me from A to Z, one of the main reasons being that I am still having fun writing code and I need it in order to stay sharp

I think that it really depends on whether the person who is building the product is technical or not. As a technical person, I'd rather prefer to have it under my control and split into pieces. In my daily job, I'm just using Claude Code under my tight control and it's proven to work really well. As a rule of thumb, I'm not fond of all-in-one tools - I have a strong feeling that every tool...

My wife is an artist and she was telling me about promoters on Instagram - you're paying them some money and they're posting stories in their account mentioning you, then people are visiting your account, which can potentially lead to increasing the number of followers (if the content on your page is good, of course). This sounds like a reasonable practice to me (even though it's robbery...

While being in early phases of a product development, I don't yet know what my final stack will be, so I'm experimenting and playing with various technologies (DBs, front-end/back-end frameworks, deployment tools etc.) and having full control over it via config files is something that is comforting me way more than having to use some CLI (I'm looking at you, aws-cli) or UI (yep, still looking...

I'd say 2-3 channels tops and a couple of occasional DMs over the day - but hey, I'm a developer, my PMs and my CEO don't need to talk to me every 10 minutes :D

Argh, I hate the AI hype (I'm actually very fond of LLMs and how they are simplifying many tasks), so I'd rather be happy to see a "No AI Inside" badge on the app, because I have a feeling that the owners of some products just have a feeling that marking their product as having AI features will sky-rocket them. Nonsense :D

I am only using the Apple Health app with an Apple Watch on my hand and... being disappointed every night before going to sleep because in most cases the only closed ring is the Stand one - I'm good at that :D

I like to re-read my posts before sending them in order to make sure that there are no typos/obvious mistakes :) Sometimes, when I'm writing a bigger post, I'm just running it through a LLM to check if it has any advice to make. Answering the question: it's indeed a gut feeling, however I'm pretty confident that sometimes I'm mistaking human posts for LLM ones :)

From a developer's perspective, the first thing I notice breaking is the feedback loop. When teams are small, what users need and what's being built stay pretty close together. As things scale, that distance grows and by the time a real user problem makes it through all the CS/PM teams to whoever is going to implement the actual feature, it's often already been interpreted into something...

Ratings, brand, UI, functionality, in that order. App Store ratings give me a quick gut check, the brand tells me if there's something real behind it, the UI tells me how much they care about the product, and then if I'm still interested I look at whether it actually does what I need. Most options fade out before I even get to step four.

I think my skills are already quietly degrading because of this. Not dramatically, but when you stop exercising a muscle regularly you lose it gradually. AI handles more of the thinking, I get the answer faster, but the mental work that used to build and maintain my knowledge is just happening less. Hard to see that as neutral in the long run.

VS Code + Claude Code in the terminal. Simple combo but it works well for a semi-manual workflow. I review everything before it goes in, so having the terminal separate from the editor actually helps keep that habit

At 25 I wanted to start a software development business, mostly for the money and the freedom I imagined came with it. Now at 37, watching my CEO, who is also a good friend, spend most of his time managing people and processes rather than building things, I realize I'm genuinely happier as a developer. I have more freedom than he does in a lot of ways, and none of the headaches that come with...

Interesting stats and a good set of questions to reflect on. On my end: database indexes are something I think about every time I add a new field or table, not as an afterthought. We do have automated tests and the rule is simple, every change ships with a test. As for where the system breaks under load, that's the harder one. We have monitoring in place that helps, but I wouldn't claim we have...

Huge fan of aluminum Macbooks with MagSafe, nevertheless this looks very promising!

Still in VS Code, still reading code the old fashioned way. Manual reviews mean I'm going through output line by line regardless of how it was generated, and for anything I actually enjoy building I'm writing it myself. Probably not the most popular workflow of 2026 but it works and I understand everything in my codebase, which feels worth something

Manual review every time, no exceptions. The one thing I've added is a dedicated review pass through Claude before I do my own, specifically asking it to look for security issues. It's a useful first filter but it doesn't replace reading the code yourself.

Fully agree with this. I've been a web developer for 15 years and most of what I learned formally never made it into actual shipped products. What did make it was knowing what you want to build, being able to tell when something is broken, and having the patience to keep going. I strongly believe that none of that requires a CS degree. The LLMs have just made that more obvious than ever!