Vasiliy Kirochka (@cromewar) header
Vasiliy Kirochka (@cromewar) profile picture

Developer advocate turned AI engineer | Going deep on reliable, production-grade AI agents, evals, context engineering | Building @cromewar in public

Joined May 2026
Photo by Vasiliy Kirochka (@cromewar)Photo by Vasiliy Kirochka (@cromewar)Photo by Vasiliy Kirochka (@cromewar)
Give us a reset, I’m like a yonkie eager to try more of my drug, but I’m out of the weekly token limit lol
Please welcome GPT-6 Sol and GPT-6 Luna to the GPT-6 universe. GPT-6 Sol and Luna build on the advances behind GPT-6 Astra, bringing much of its strengths into faster and more affordable models to support work at scale. We’ve also made caching and inference more efficient, and we’re passing the savings directly to you: 50% lower API prices for Sol and Luna compared with GPT‑5.6 promotional pricing.
18
Multi billon dollar company can only create this crap while closing fan projects that have multiple times more passion and understanding of their IP. No wonder why they closed projects like Ascension, they were exposing how bad blizzard is…
WoW: Forever launches November 4 with: ⛓️ New dungeons 👨‍👨‍👦‍👦 10 & 20-player raids 🪽 Skyborne playable race 🗺️ New & expanded zones ✅ Account-wide perks 💡 Updated world visuals & lighting 🧟 New race & class combos ❗ 1000+ new quests Join the Beta September 17!
34
It’s a good start, but competitors have no incentive to slow down and pacing yourself alone only means surrender. This must be a worldwide consensus where China and the US pace the development, there is no other way.
We Must Pace the Frontier: I’ve written a new essay on why the AI industry should slow down, with a three-part plan for doing so. Anthropic is unilaterally committing to the first of these steps. We’ll provide third-party evaluators with permanent, employee-level access to our systems, so that they can verify adherence to our safety measures, report on incidents, and assess models’ alignment during training. You can read the full post here: darioamodei.com/post/we-must…
14
Crazy times ahead!
We’re sharing a solution to the Navier-Stokes Millennium Prize Problem, one of the deepest problems at the frontier of mathematics. The proof was produced by a group of agents, using an OpenAI next-generation model significantly more capable than GPT-6 Astra. The problem concerns whether the description of smooth three-dimensional fluid motion modeled by the Navier-Stokes equations can break down. It has remained unresolved for roughly 90 years.
16
I built a Pomodoro widget for @kde Plasma 6. You type what you're working on and it stays on the panel next to the countdown. Presets, estimated finish time, survives a plasmashell restart. MIT. github.com/cromewar/kde-pomo…
1
46
The whole point is dragging a file straight into another app, so I tested that myself, and it works. I don't take the AI's word that something is done, I check it.
7
The bugs that cost the most were boring framework trivia, not logic. Qt 6 had quietly renamed a property, and a Plasma widget won't render unless you define one specific thing. Both cost hours.
16
The honest split: the AI wrote the code, I drove the direction, the taste, and the testing. I didn't write the fix for a single hard bug. I decided what was wrong and where to look.
13
I'm writing this down publicly os it's harder to back out of. For the next 6 months I'm going all in on AI engineering and building in public. I'm at the start of this, and I'll be honest about where I actually am the whole way.
4
1
21
So that's the deal I'm making with myself: show up, build, and tell the truth about how it's going. Come back in 6 months and we'll see what happened.
6
Where I'm actually starting: @edwarddonner LLM Engineering course, then a small public project every couple of weeks, writing down what I learn and what I screw up along the way.
10
Here's the path I picked. - Applied AI first: agents, RAG, prompting, evals. - The heavy ML and math can wait, for now. - Build small things and actually ship them, not just collect courses.
12
Nobody's really watching yet, and that's fine. This isn't for an audience. It's me putting a stake in the ground so future me can't pretend I didn't try.
12