Zachary Schneider ☁️

Narrative violation and great insight from the latest Citadel Securities banger by Frank Flight: "We illustrated back in February that demand for software engineers, the most AI exposed occupation was accelerating higher, which we argued violates the displacement narrative. Indeed the acceleration in software job postings has continued, now up 18% from the inflection point in May last year."

oh you're a cannon expert? name all the cannons

. @tensorlake sandboxes now support GUIs for enabling computer-use agents. Spin up a sandbox with VNC + XFCE pre-installed so the agent can interact with the GUI using tool calls. You can also hook into the VNC stream from your apps and record sessions in real time. Sandboxes support fully dynamic resource allocation, so you can configure the right amount of CPU, memory, and disk based on what the agent needs to do. The harness code is open-source. We also explain how the harness works internally, and how it interacts with the sandbox and tunneling infrastructure, in the links below 🧵

And code.

Now that PDT—the only thing strangling retail investors—is terminated. I’m publishing a repository explaining everything in extreme detail. 1. How SMA outfits manufacture precise signal generation—often called ‘buy algorithms’ 2. How the Major Market’s [S&P 500, Nasdaq, Dow, and VIX] systems operate midterm trends with itemized information 3. How SMA outfits dual-program, or coexecute, on inverse vehicles 4. How financial institutions weigh capital-risk when positioning based on: - where each System is trading ( + or - ) - ranked outfitting 5. Cryptocurrency And surmising how the above information inherently manifests as an unfair market. Thank you!

Deno 2.7.14 is out with 120+ bug fixes Major improvements and fixes to node:http2, node:tls, node:net, node:dns, node:cluster, node:crypto, node:timers, node:stream, Web API, and more github.com/denoland/deno/… See you next at Deno v2.8

Basejump is live on @hydration_net Bridge EURC from Base to Hydration in 22s ⚡️ Instant bridging experience that eliminates user friction Up to 10k EURC at a time Try it: app.hydration.net/cross-chain?sr…

A Rust dev just killed Headless Chrome. It's called Obscura. The open-source headless browser purpose-built for AI agents and scrapers at scale. Chrome vs Obscura: - Memory: 200MB+ → 30MB - Binary: 300MB+ → 70MB - Page load: 500ms → 85ms - Startup: 2s → Instant - Anti-detect: None → Built-in Single binary. No Node, no Chrome, no dependencies. Stealth mode is brutal: → Per-session fingerprint randomization (GPU, canvas, audio, battery) → 3,520 tracker domains blocked by default → navigator.webdriver masked to match real Chrome → Native function masking so detectors can't sniff it out Drop-in replacement for Puppeteer and Playwright over CDP. Zero code changes. If you run agents or serious scraping at scale, this repo prints money. 100% Opensource.

Wireguard is under appreciated for building overlays.

Long gold is a better expression of "policy is easing into resilient growth" than long bonds. Bonds need a growth scare or a Fed pivot to rerate higher. Gold needs neither. It only needs real rates to drift lower and the dollar to fade, both of which are base case I explained this in the livestream linked below

What if your whole computer were just pixels streamed to you from a model? I’ve been working with @zan2434 and @drewocarr to imagine a version of generative computing that’s much more flexible and visually rich than the GUIs we have today. (Video is sped up and edited)

what are we getting now? operating systems which request proof of age more mass surveillance mass surveillance companies trying to be cool kids more facial recognition more internet, chat, e2e, and phone lockdowns idk seems like we need something new to fight all this

where i'm at on LLMs: 1. LLMs are NOT going to outmode software engineers, not now, not ever 2. LLMs alone are a dead end for attaining anything like AGI 3. forget about LLMs achieving ASI 4. LLMs will only make good engineers faster, marginally 5. LLMs will remain poor at architecture and design this is 100% because LLMs cannot reason and no, generating a bunch of hidden context is not "reasoning" as always, when the same people are making claims like this, have a look at what they stand to gain and to them, i warn directly: you are going to fall. hard

Solr isn’t "losing". It’s being abandoned. Not because #ApacheSolr is worse than anything, but because no one remembers why it’s still there. The pattern is always the same: • The Solr expert left 18 months ago • Apache ZooKeeper is "working" (so nobody touches it) • Everything else is Kubernetes-native… except Solr By the time migration comes up, the decision’s already made. The real question isn’t should we move It’s: how do we migrate without breaking relevance, latency, or the roadmap? If you’re in that quiet evaluation phase - skip the vendor fluff. The real checklist is shorter than you think.

Introducing OpenMythos An open-source, first-principles theoretical reconstruction of Claude Mythos, implemented in PyTorch. The architecture instantiates a looped transformer with a Mixture-of-Experts (MoE) routing mechanism, enabling iterative depth via weight sharing and conditional computation across experts. My implementation explores the hypothesis that recursive application of a fixed parameterized block, coupled with sparse expert activation, can yield improved efficiency–performance tradeoffs and emergent multi-step reasoning. Learn more ⬇️🧵

@SmallCapBob2 Imagine a Trump ticker tweet of $GME right now ...

With the Straight of Hormuz now open for the remainder of the cease-fire, Donald Trump just provided Ryan Cohen & #GameStop the perfect opportunity to announce their MEGA/MERGER as the S&P reaches an ATH THIS TWEET IS RYAN’S WAY OF THANKING HIM. $GME $SPY

If you think Private Credit is terrible for investors, check out Fund of Hedge Funds products from major bank wealth managers with billions in AUM. The client gets charged 670bps to enjoy a 3.4% return over cash over the last few decades.

Apple Silicon + Gemma 4 fans: this is for you. Pico AI Server now supports continuous batching with MLX-Swift. 43 tok/s on 1 stream. 26 tok/s per stream on 2 concurrent streams. That’s 52 tok/s total. a 21% throughput gain on a six-year-old MacBook Pro M1 Max!

10BN to buy.

Peter Mandelson failed UK Foreign Office vetting ft.trib.al/L66WDf7