lately i have doing small implementation of casandra paxos lwts, tigerbeetle viewstamped replication, cockroachdb raft per replica group playing around with different consensus ideas

Running evals against deleting my entire skills / docs and yoloing with the frontier agents and no skills. Early returns: this is a terrible idea

AHAHAHAHAHAHAHAHAHAHAHA NO WAY. > OpenAI just announced Dots. > SpaceXAI bought Dot.com it redirects straight to the Grok Bot download page. based lmao.

everything is fake - posts on here, ghostwritten - statistics, misleading - benchmarks, bullshit - reviews, paid for nearly every company is doing this. the frenzy over ai is showing me so few founders have any backbone

Wild, even Cursor moved away from server-side RAG index and into Grep-like agent tool calls for code search "we’re no longer computing embeddings of your code or storing them on our servers for search." — Jul 20 huh

100% of people take the first offer they get out of school, defining their future for all of eternity, since 0% of people ever change the direction of their career how can you leave the most important life choice to chance, just to avoid the small discomfort of interviewing more

we got something exciting for you 👀

🚀 Research Opportunities in Multimodal AI for Healthcare, IIITH Our 3-year research project under ANRF MAHA AI for Science is kicking off, and we are building a research team! We have multiple openings for JRFs and Postdoc working at the intersection of AI and healthcare.

Periodic sadness reminder that the anthropic subscription still does not permit third party harnesses :(

Most people view themselves more negatively than the world does. You see 100% of yourself all the time, all your strengths and flaws, while seeing everyone as they present themselves. No one tries to present themselves badly, so you're most often only going to see the good parts of everyone else. Which means you, seeing both your good and bad and only everyone else's good, think you're worse than everyone else. You're innately biased to the negative.

In retrospect, many of my largest life decisions can be traced back to a single YouTube video popping up in my recommendations sidebar, or a book that I picked up by coincidence because it had an eye-catching cover. I'm still grappling with the idea that my laptop gives me the option to shape my personality to a terrifying degree.

With networking in general the most important time to interact and build relationships is when you do not need anything. By the time you want someone's support or you want someone to care about what you've done it is way too late to start building a relationship.

aws made the industry soft. we're used to them handling all the hard problems and everyone can be a pure product company with 85% margins ai infra is not mature enough for this but everyone is pretending like they can offload the actual work to someone else it's why there's 27 model router products. even most inference companies do the bare minimum on compute someone else set up and it all feels successful because there's so much money around it's hard to justify doing the hard thing when the easy thing is profitable but the gap between what openai, anthropic, deepseek, moonshot and everyone else is huge and it's easy to forget that someone actually has to do the work to close it

What happens as the cost of creating software goes to zero?

I am actually getting 99.7% cache hits with DeepSeek Harness and a self-hosted GLM-5.3 This is the difference between vibe coded harnesses and quality engineered ones

Most predictions I see are still way too conservative. Here's mine

I built the same small macOS app in Obj-C + AppKit, Swift + AppKit, SwiftUI, React Native macOS, Tauri, and Electron. Observations: 1. Obj-C or Swift + AppKit are tiny in binary & mem, TTI 2. RN macOS has a good TTI, decent memory 3. Tauri's TTI was worst, but smaller bundle than RN 4. Electron is massive in nearly every way Not shown: how fast it reloads if you make a change. React Native, Tauri, and Electron win that one -- RN fastest. I enjoyed making the Objective-C one the most. Of course, basing your entire decision on just this chart would be incredibly dumb and tunnel-visioned. There's much more to tech decisions than just binary size, memory footprint, and TTI. For example -- do you need to support more than just macOS? But it was a fun exercise! (Not open source, because this is actually an internal macOS app we might use at Infinite Red for some ... stuff. Feel free to do your own tests, but do them right, like make sure to strip symbols and don't drag in unnecessary dependencies and whatnot!)

People rewrite in Rust or Go and claim that as the reason for huge performance gains, at face value that's cool but when looking at what actually makes stuff faster it's almost never the language. A small thread.

OpenCode Goon

The "whiteboard defense:" I should be able to pull you aside at any moment and ask you to explain any customer-facing system you've shipped. You should be able to clearly explain how it works and defend the decisions you made. This is my benchmark for responsible AI usage. I don't expect line-level familiarity with the code. I don't care if you remember the exact function name or implementation detail. You may not even know it. I don't care. But if I ask "why did you do X instead of Y?", "what happens if this actor behaves maliciously?", "what data structure did you use here and why?", or "where does this fail?" you should be able to answer confidently. For PoCs, demos, experiments, whatever: I don't care. Generate 100% of it and understand none of it. Speed over quality every time in those specific scenarios. But if you're shipping customer-facing work, you can't be shipping things you don't understand at a high level.