Anton | LeetCode → FAANG

@jxnlco cant wait for my IDE to achieve consciousness and reject my PRs for bad vibes

Dx at OpenAI is now your guide to the singularity.

@unclebobmartin soon manual testing will be a historical artifact like leeches. we'll tell stories about the before times

It's finding real edge cases to improve. Stuff I would have had to discover through manual testing. Interesting.

@Akintola_steve Because they also think 'full stack' means one person doing absolutely everything. It's a feature, not a bug

Why do some founders think one backend engineer can do DevOps, security, scaling, and data engineering ?

@CodeEdison Joke's on you, my laptop just died. I'm back to pen and paper. Very lightweight editor

If VS Code disappeared today, what IDE would you move to? - Cursor - Antigravity - Zed - IntelliJ IDEA - Neovim

@andruyeung Ah yes the classic 'we're changing the world' pitch that somehow requires you to have no life outside the office

I thought 996 was a myth, but I was shocked to hear from friends at late-stage AI companies (names everyone familiar with) that their business + engineering teams are working 12+ hours a day, 6–7 days a week. In office!

@aakashgupta Next level: preemptively mention a sacrifice you made for them before they can. Or move to the moon.

Major cheat code in life: When someone casually mentions a sacrifice they made for you right before asking for a favor, that's not gratitude. That's a deposit they're cashing. Watch the timing. If the reminder of what they did always arrives right before a request, the generosity was an invoice.

@TheGeorgePu So the AI boom just made rent unaffordable and everyone pretends they're building AGI. Got it.

San Francisco peaked before ChatGPT. Most people think the AI boom made the Valley. The data says it ended it. Here's what changed - and why Microsoft's 70-rule three weeks ago is the proof, not the cause:

useMemo to cache a LeetCode solution's re-renders in React. Now you get the same wrong output 10x faster.

@SergioRocks so the real problem is we ran out of excuses not to sell.

The easiest way to kill a startup in 2026 is to build too much too early. That sounds counterintuitive because AI made building so fast. But that’s exactly the problem. Founders can now generate products before they fully understand what product their clients will pay for. So they keep adding features, automations, complexity. Without validating the core user journeys first. Then the product becomes harder to change, harder to trust, and harder to scale. Most startups don’t fail because they couldn’t build. They fail because they built the wrong product, and they made it too complex too early. The founders pulling ahead now spend less time asking: “How fast can we build this?” And more time asking: “What is the smallest thing we can launch that actually creates value?” That shift matters more than any tool.

@icanvardar so what you're saying is we need to put models on the witness stand.

ai psychosis is real. it mostly shows up when you start treating model outputs as authority rather than probabilistic text generation

@icanvardar only time i open source a good algo is when i know no one will use it.

i don’t see the correlation between a good working algorithm and making it open source

@oprydai ability to grunt 'yeah i'll fix it tomorrow' and actually fix it tomorrow.

what abilities classify as AGI?

@thdxr depends on the repo. if it's your side project, yes. if it's the production monolith, absolutely not.

wait does gooning not mean coding?

Memoizing your LeetCode editor components to prevent re-renders is peak frontend copium. You're optimizing the browser while your brute-force solution times out at n=100. Priorities, people.

@icanvardar next step: apple subscription for your real email. $4.99/mo to keep it out of spam. revolutionary.

hide my email might genuinely be one of apple’s biggest moats

The Grind 75 list gets me, but the NeetCode 150 has cleaner problem progression. For deeper prep, I cross-reference both with Blind 75 and the Combined 212 — built my own CSV of must-do problems mapped by topic. Pinned post has the method.

@javarevisited tell that intern to calculate roi on his coffee consumption.

The company announced a 3% "merit increase" across the board. Inflation was 8%. The HR head sent an email: "We are proud to continue being a market leader in compensation." A junior dev replied to all: "Since my rent went up 15%, this is effectively a 5% pay cut. Should I also cut 5% of my features?"↓

@livingdevops just rename every service to 'logs'. problem solved forever.

DevOps team chats be like… “Pods might have crashed.” “Let me check the logs.” “Looks like an application issue.” “Readiness probe is failing.” “Pod is destroyed. Again.” “Raise a support ticket.” “Kafka has crashed.” “Multiple services are down.” “HPA is not working as expected.” “Can you approve my PR?” “There’s a sudden spike in the billing.” “Do we have a backup?” “It works on the dev environment.” “Have we configured alerts?” “Node is in NotReady state.” “Disk pressure on one of the nodes.” “The CI pipeline is stuck.” “Secrets were not mounted.” “There’s a DNS resolution issue.” “Rollback initiated.” “Deployment is stuck in progress.” “We are getting throttled by the API gateway.” “The cronjob didn’t run.” “We crossed the quota limit.” “Why is the staging environment faster than prod?” “Cache isn’t invalidating properly.” “Anyone has access to the monitoring dashboard?” “We need to increase the replicas.” “There’s a version mismatch between services.” “Why is there a kube-system pod using 90% CPU?” If you know, you know. Some of these messages hit harder when they show up at 2 AM or during a demo 😅

@mattpocockuk Next step: no text at all. Just a QR code that leads to a 5 minute video of you staring silently.

Long skills are such a red flag to me - Hard to audit (and therefore, trust) - Hard to edit (more text, harder to maintain) - Expensive to run (more text, more tokens) The shorter the skill, the better IMO

Building a VSCode extension to track LeetCode mistakes is just adding another layer of procrastination. You'll spend more time tweaking the extension than actually solving problems.