stengineer

couldn’t figure out where everyone gets their linkedin post gifs so i made one myself in @cavalry__app 🤲 slowly learning the tool by recreating falloff type effect by @algotv 🐐

Accelerating and automating science and research is one of the noblest pursuits right now. We need to jointly train not just single meaning units like word vectors, not just embed all sentences, not only train one model to be prompted by any question, but ideally the entire scientific endeavor - the meta cognitive layer of science and its communities. This paper is a useful step towards this. This type of acceleration is also exemplary for why computer science and AI move faster than many other fields.

Brut-V is a low-level graphics runtime written in pure RISC-V assembly, designed for sketching and learning at the instruction-set level. It was built entirely with Hermes as part of @NousResearch Creative Hackathon. The agent also helped automate the creation of surprisingly convincing visuals around it. A few notes from the build (+ live demo) below

Features - Multi-agent: each task runs on a specialized profile, with its own tools, skills, and personality. - Linked tasks: parent → child dependencies. Fan out work, gather results, continue. - Shared workspaces: agents hand off files through a directory, a git worktree, or a temp scratch space. - Live dashboard: watch the team work in real time. - Comments per task: humans and agents both write. The next agent picking up the task reads the full thread. - Skills built in: a planner (kanban-orchestrator) and a worker (kanban-worker) ship by default. Pin a domain skill per task to load expertise. - Durable: SQLite-backed. Survives crashes, restarts, and reboots. - No double-claims: when many agents compete for the next task, only one wins. - Heartbeats + runtime caps: long jobs stay alive, runaways get killed. - Project isolation: run multiple projects on one machine without crosstalk. github.com/NousResearch/h…

Hermes on Kanban — nice! ✨ Btw it also runs in our open-source Kanban alongside Claude Code, Codex, Devin & more. 🔥 Try open-source Claude Design → github.com/nexu-io/open-d…

A Lens That Takes Derivatives US Patent Basis: US8610839B2 - Optical Processing System for Computing Derivatives. In a 4f Optical Processor, the first lens takes the incoming field and forms its Fourier spectrum. At that middle plane, a tiny optical mask multiplies the spectrum by iξ. This is the derivative operator written in Fourier language u(x) -> U(ξ) -> iξU(ξ) -> ∂u/∂x Then the second lens brings the field back to the real space. What comes out is no longer just a focused beam. It is the spatial derivative of the input field, computed by light as it propagates. So, this is the serious promise of optical computing. A physical optical train can perform operations that usually live inside numerical code: differentiation, filtering, convolution, edge detection, correlation, and many other linear transforms.

We implemented @karpathy 's MicroGPT fully on FPGA fabric. No GPU. No PyTorch. No CPU inference loop. Just a transformer burned into hardware, generating 50,000+ tokens/sec. The model is small, but the idea is not: inference does not have to live only in software 👇

JP Morgan's investment research team just shared exactly how they built their multi-agent system "Ask David", and it's the same architecture pattern showing up everywhere: - supervisor agent orchestrates - specialized subagents handle retrieval, structured data, analytics - LLM-as-judge reflection node before the answer ships - human-in-the-loop for the last accuracy gap worth watching for anyone building:

another beautifully designed website curating digital, physical artifacts of tech 🌐 by @p0 museum.parallel.ai

I used Hermes and Kimi to build an agentic workflow for creating custom typefaces. @NousResearch @Kimi_Moonshot A visual interface works alongside Hermes chat to facilitate collaboration between agent and designer. The designer provides two hand-drawn sample glyphs, which Hermes uses to generate the full typeface in the target style, requesting feedback along the way.

Machine Telephone ☎️ Enter a concept. 'Dial a machine' to pass information across different modalities and see how context shifts or evolves as it's interpreted sequentially.

I've started writing my book: "Biology is a Burrito & Other Essays." It is an interactive and highly visual look into the beauty, speed, and complexity of a living cell. I'm planning to print hardcover books while serializing the essays online. The first essay is now available at burrito.bio. This was inspired by Stewart Brand's latest book, "Maintenance of Everything," which he developed in serialized form with @WorksInProgMag. One cool thing about that book was that he improved each chapter with reader comments before printing the physical copies! I'll be doing the same with this book. If you send me feedback that improves the text, I'll credit you online and in the final print version. You can also sign up to get email updates when a new essay launches. Hope you enjoy!

please please inject it into my veins @rybinfx

Taught Hermes how to use pretext 😊 @_chenglou @NousResearch

views from the ocean floor

fih

I made SpaceTimeLoom.com to explain gravity to my kids. They can just play with mass and see how it bends space. You don’t have to explain concepts anymore. You can build them!

Engineers at OpenAI experience the same problem as everyone else — we can supervise about 3–5 coding agents. After that productivity drops. Codex is smart, but our attention is limited. So we built (and open sourced!) Symphony to remove that ceiling. Here’s how it works:

時の流れがゆらぐように感じる、最高に美しい物理表現のインスタレーション！we+と島津製作所の『In-Between Matter』。分析技術の要となる「分離」を、視覚で体感できる装置として昇華させたもの。見る人の科学への関心を誘う、素敵なアプローチ。 #MilanDesignWeek2026 #ミラノデザインウィーク2026

The top trending repo in the world is mine Madness