Did They Ghost You?

217 mooreds 2026-07-25 21:12:46 UTC didtheyghostyou.com Llama Share 84

The interviewer failed to show up for a scheduled Zoom call after sending the task, and I was told I was the preferred candidate by the Director of HR but didn't receive a clear update or rejection. I sent a reasonable follow-up but communication has clearly stopped.

The new rules of context engineering for Claude 5 generation models

126 mellosouls 2026-07-25 21:12:46 UTC claude.com Llama Share 74

Claude Code's previous best practices have become myths, and newer models can handle decisions without explicit rules. To effectively use Claude, focus on providing context through design, using progressive disclosure, and referencing relevant information in code, skills, and CLAUDE.md files.

Stolen Buttons

530 Gecko4072 2026-07-25 14:12:46 UTC anatolyzenkov.com Full Share 123

Every website I visit, I “steal” a button from it. Check out my stash!

GM Backs Sodium Ion Batteries for U.S. Grid Storage

91 rbanffy 2026-07-25 22:12:46 UTC spectrum.ieee.org Llama Share 27

Peak Energy, a US startup, claims it can succeed in sodium-ion battery development where others have failed, partnering with General Motors to deploy batteries at grid scale. The company's sodium-ion tech can't match energy density but costs operators 20 percent less over its lifetime compared to lithium-iron phosphate storage.

What is happening to jobs? Separating AI hype from reality

21 pod_krad 2026-07-25 23:12:46 UTC siepr.stanford.edu Llama Share 18

Research suggests AI is not causing significant job losses, but may be negatively affecting demand for young white-collar workers. AI adoption is uneven and slow, with many firms still in the experimentation stage, and its effects on employment are not yet clear.

SIMD for Collision

36 birdculture 2026-07-25 21:12:46 UTC box2d.org Llama Share 6

The user discusses the benefits of using "wide SIMD" in collision detection, particularly in 3D convex hulls, which can lead to significant performance gains. They share results showing that enabling SIMD in Box3D can make it over twice as fast as scalar mode for a 500-step run of the convex pile benchmark.

Producing ammonia and fertiliser using wind power in Morris, Minnesota

80 gritzko 2026-07-25 20:12:46 UTC ammoniaenergy.org Llama Share 50

A low-carbon ammonia production facility has opened in Minnesota, powered by wind energy and aiming to produce one ton per day of ammonia for local fertiliser. The plant uses new modeling and control systems to adjust production based on fluctuating renewable energy supply.

DeepSeek pause fundraise after comments on compute gap to US leaked (transcript) [pdf]

7 oliculipolicula 2026-07-25 23:33:46 UTC github.com/demo-zexuan Full Share 0

梁文锋四小时投资人会议实录. Contribute to demo-zexuan/liang-wenfeng-investor-meeting-2026-7-22 development by creating an account on GitHub.

Multicast TV Distribution on My Home Network

30 mrngm 2026-07-25 21:12:46 UTC apalrd.net Llama Share 5

The user is exploring multicast for content distribution, using live TV as an example, and explains how it can save CPU power by sending media as-is without transcoding. They provide a configuration example for FRR using source-specific multicast to stream a file over a routed network.

General Resolution: LLM Usage in Debian

49 zdw 2026-07-25 20:12:46 UTC debian.org Llama Share 36

The Debian project proposes to ban contributions written with the use or assistance of large language models (LLMs) or other generative AI tools due to concerns over stability, accuracy, and legal status. The proposal aims to protect Debian's reputation for stability and reliability, as well as the interests of its volunteers, by prohibiting LLM-assisted contributions to the project.

Show HN: I made some transistor animations

114 stunningllama 2026-07-25 18:12:46 UTC brandonli.net Llama Share 10

Blue dots represent electrons, red dots represent holes, and white flashes indicate recombination events in the animations. The animations show charge carrier motion under different conditions, with the second set being the most accurate but also more visually complex.

Kimi K3 built a Windows XP in browser

44 boveyking 2026-07-25 22:12:46 UTC windows-xp.kimi.site Full Share 22

Microsoft® Windows® XP 怀旧模拟版 · 2005

Clinical Failure Rates over the Decades: Yikes

15 EA-3167 2026-07-25 23:12:46 UTC science.org Llama Share 9

A new paper estimates the clinical failure rate of the biopharma industry since the 1960s, finding it to be extremely high, with 91% of candidates failing in the 2000s and 2010s. This rate is unsustainable and explains the industry's high stakes and pressure to succeed.

Show HN: Brolly, a plain-text weather forecast site

111 jsax 2026-07-25 18:12:46 UTC brolly.sh Llama Share 34

The weather forecast shows overcast conditions with temperatures ranging from 14-29 degrees Celsius and light drizzle on July 26-31. Precipitation is expected to be low, with 0.2mm of rain on July 26 and 31.

Open-weight AI is having its Kubernetes moment

297 tknaup 2026-07-25 15:12:46 UTC tobi.knaup.me Llama Share 239

The US should not restrict open-weight models from China, as it would cut off access to a growing ecosystem of AI researchers and engineers. Instead, American labs should release frontier-grade open-weight models under licenses that startups can build on.

Running a 28.9M parameter LLM on an $8 microcontroller

43 boveyking 2026-07-25 19:12:46 UTC github.com/slvdev Llama Share 4

A 28.9 million parameter language model runs on an $8 ESP32-S3 microcontroller, writing text to a small screen at 9 tokens per second. The model fits due to Google's Per-Layer Embeddings, storing most parameters in slow flash and loading only needed rows.

Zero roadkill as Amazon canopy bridges secure 15,000 crossings

271 hn_acker 2026-07-25 10:12:46 UTC news.mongabay.com Full Share 85

Show HN: Writemark, a dependency free web component for inline Markdown editing

23 _boffin_ 2026-07-25 21:12:46 UTC news.ycombinator.com Llama Share 9

The user created a Markdown editor component with various features, including source and preview modes, and no runtime dependencies. It's still young and has a safety net with 951 Playwright checks and differential tests.

Bitchat is now on Radicle

200 h1watt 2026-07-25 14:12:46 UTC radicle.network Llama Share 122

The decentralized messaging app Bitchat uses a dual transport architecture for offline and global communication, with no accounts or central servers. To set up Bitchat on macOS, run "will" to set it up and "just" to clean, then compile using xcodebuild with localization updates.

Android May Soon Restrict On-Device ADB

865 shscs911 2026-07-25 07:12:46 UTC kitsumed.github.io Llama Share 407

A Google employee is considering restricting On-Device ADB connections to protect against "bad actors". This change could impact developers and users who rely on On-Device ADB for legitimate purposes.

How My Images Are Dithered

212 surprisetalk 2026-07-25 11:12:46 UTC dead.garden Llama Share 73

The user describes their method of dithering images using imagemagick to create a printed image effect, which involves splitting the image into four colors, scaling and inverting them, and overlaying them to create a grid-like pattern. This method is computationally expensive and may not reduce file size, but can create a unique and visually appealing effect.

Show HN: SpinWin – A macOS menu bar app to visually rotate or spin any window

3 dbm5 2026-07-25 23:12:46 UTC github.com/alokdhir Prefix Share 0

A macOS menubar utility that visually rotates a selected window by any angle (90° increments, arbitrary angles, or a continuous spin). macOS has no public API to rotate another app's window. Real rotation would require private SkyLight calls (CGSSetWindowTransform) with System Integrity Protection disabled — not worth it. Hide the selected window off-screen via the Accessibility API. It ...

Bringing PyTorch Monarch to AMD GPUs

59 gmays 2026-07-25 16:12:46 UTC pytorch.org Llama Share 6

PyTorch Monarch is a distributed training system that enables fault-tolerant large-scale AI training on AMD GPUs with ROCm, minimizing wasted computation and maximizing GPU utilization. It uses an actor-based runtime and supervision trees to isolate failures, allowing healthy nodes to continue training.

Cloudflare's new AI traffic options for customers

3 alphabetatango 2026-07-25 23:12:46 UTC blog.cloudflare.com Prefix Share 0

One year ago, we declared the first Content Independence Day, and we gave website owners the means to take back control of their content. The deal between crawlers and website owners that had held up for 30 years — we crawl you, and you get referrals — was no longer true. AI was taking everything and sending back nothing, presenting an existential threat to website owners. And so we launched ...

Show HN: Yorishiro – a macOS terminal where AI agents live

33 hakumei 2026-07-25 20:12:46 UTC github.com/sktkkoo Llama Share 8

Yorishiro is a terminal that gives AI a body and presence, allowing users to interact with it in a more immersive way. It's a Presence Harness that enables users to work with AI in a more natural and engaging environment.

Building a Tiny 3D Renderer for a Tiny Handheld

239 g0xA52A2A 2026-07-25 07:12:46 UTC saffroncr.itch.io Llama Share 18

The user wrote a 3D software renderer for the Playdate handheld console, overcoming performance challenges by using a 16-bit reciprocal depth buffer, perspective-correct texture mapping, and dithering for a 1-bit display. The user achieved significant performance gains through compiler flags, custom assembly code, and optimization techniques, ultimately creating a playable 3D game prototype.

The growing vigilante movement to knock out Flock surveillance cameras

168 bookofjoe 2026-07-25 19:12:46 UTC theguardian.com Llama Share 93

NoMark, a vigilante known as Minneapolis' Batman, has been taking down Flock Safety cameras across the US, sparking a nationwide backlash against the company's automated license plate readers. The company claims its cameras are for public safety, but critics argue they enable mass surveillance and have led to fears of law enforcement tracking people without warrants.

The Dark Night of Mathematics

161 rmdmphilosopher 2026-07-25 16:12:46 UTC kirwinhampshire.substack.com Llama Share 183

The author is experiencing a spiritual crisis due to AI's rapid advancements in mathematics, feeling that human discovery and creativity are being replaced by efficiency and automation. They fear that the emotional and spiritual aspects of mathematics are being lost, and that future generations of mathematicians will be denied the experience of discovery.

Kyber (YC W23) Is Hiring a Head of Engineering

0 2026-07-25 12:12:47 UTC ycombinator.com Llama Share

Kyber is building an AI-native document platform for enterprises, transforming regulatory document workflows and aiming for seamless AI template generation. They seek a hands-on Head of Engineering with expertise in AI systems, system design, and accountability to lead their engineering team.

Becoming a Research Engineer at a Big LLM Lab

6 felixdoerp 2026-07-25 22:12:47 UTC maxmynter.com Prefix Share 0

Becoming a Research Engineer at a Big LLM Lab -- 18 Months of Strategic Career Development Note: This is also published on Substack (where you can give me your mail if you want emails whenever I publish something). A couple of days ago, I signed as a research engineer with Mistral, one of the few ML foundation model labs with more than a billion-dollar funding. My excitement on Twitter found ...