David

Frontend/Fullstack AI Engineer with 8 years of xp @ in React, React Native, Node.js and other things

[Neural Space @NeuralSpaceIO]

🩻⚡ CE NE SONT PAS DEUX PATIENTS. C'EST LE MÊME, À QUELQUES SEMAINES D'INTERVALLE.

À gauche, un corps envahi. Chaque tache noire est une zone où le cancer dévore du sucre à toute vitesse.

C'est le principe du TEP-scan : on injecte un glucose radioactif, et les cellules les plus voraces s'allument. Les tumeurs, elles, sont insatiables. Elles se dessinent d'elles-mêmes.

À droite, presque plus rien.

Entre les deux images, une seule intervention. On a prélevé les globules blancs du patient. On les a reprogrammés génétiquement pour qu'ils reconnaissent enfin ce qu'ils ne voyaient pas. On les a multipliés par centaines de millions. Puis on les a réinjectés.

C'est la thérapie CAR-T.

Et ce n'est pas un médicament au sens habituel. Ça ne se dissout pas, ça ne s'élimine pas. Ce sont des cellules vivantes qui continuent de se diviser dans le corps et de chasser toutes seules.

Le patient devient son propre traitement.

[Jeremy Wick @jecwick]

I'd rather go to jail than add a cookie banner to my website

[Ole Lehmann @itsolelehmann]

i think AI is about to uncover an entire hidden world of animal communication

and what it's already found is insane...

> AI found evidence that elephants call each other by name. when researchers played one its name-like call, it approached the speaker faster and called back more, revealing a form of personal naming once thought almost uniquely human

> AI analyzed 53,993 marmoset calls and found tiny monkeys may use names too. entire families use similar labels for the same monkey, suggesting they learn names and dialects from each other

> AI learned to tell what Egyptian fruit bats were arguing about: food, mating, sleeping spots or personal space. it could even partly identify who was shouting at whom inside a colony of thousands

> AI discovered a “phonetic alphabet” in sperm whale clicks, with rhythm, tempo and tiny timing changes combining into at least 143 recurring patterns. their communication system is nearly 10x more expressive than scientists previously believed, and we still don't know what most of it means

> an AI trained on 1.5 million zebra finch calls held real-time vocal exchanges with living birds, generating calls as they spoke and getting them to respond with the same timing and flexibility they use with other birds

> Google trained an AI on decades of dolphin recordings to predict what sound comes next, then paired it with an underwater device that gives objects their own synthetic whistles. the goal is for a wild dolphin to copy a whistle to ask a human for a specific object, creating a tiny shared vocabulary between species

> AI combined microphones worn by wild crows with nest cameras and uncovered quiet calls that may announce when a crow is arriving at the nest and help entire families coordinate caring for their chicks

> a robot bee performed the waggle dance real bees use to share the direction and distance of food, and the bees changed their flight paths based on its instructions. researchers effectively sent a destination into a hive in bee language

[Konrad Reczko @reczko_konrad]

Depth-aware light injection in TypeGPU

I got a 448x448 monocular depth model down to ~8 ms on my M4 Pro across ~250 dispatches, which is fast enough to use in realtime :D
Since the inference is written directly in TypeGPU, I can just feed the depth buffer straight into the lighting pass. It never has to leave the GPU or go through any extra synchronization/interop step

Inference, lighting and draw all go through the same command encoder.

[Stéphane - smo @_smontlouis]

Je suis devenu (inutilement) EXTRA niveau design avec l'avènement des agents. J'ai jamais eu le temps de pousser l'UI/UX, un domaine qui m'a toujours passionné donc maintenant je m'amuse à créer tout ce qui me passe par la tête.

Par exemple j'ai voulu refaire le onboarding de ma petite app biblique bible strong ce week-end:

Idéation sur https://t.co/ydjVmAvsyZ (voir les petits agents qui te pondent du design c'est une expérience incroyable!),

Je passe tout ça à codex (les png et les json) et avec quelques itérations il m'a pondu mon onboarding.

Je lui ai même demandé de me faire un "moteur" de onboarding qui me permet de définir mes scènes et avec un id les composants persistent / fade entre plusieurs scènes. Du coup j'ai mon petit moteur déclaratif hyper simple avec un système de nodes, de connecteurs, et d'objets draggables.

J'ai voulu tenter la suppression du bouton "Continuer". Laisser l'utilisateur découvrir l'app en interagissant avec l'onboarding par des hints colorés, je trouve ça plus immersif perso. S'il veut il peut toujours skip avec le bouton. Je sais pas si l'user va aimer, mais je suis satisfait en tout cas !

Auto critique:
- Beaucoup de texte dans les illustrations, ça peut être un peu trop verbeux et manquer de simplicité

react-native + reanimated, rien de fifou

[David @daavidaviid]

@jackfriks Same here 🙌

[Michael Jaunasse @qulbut0ke]

Zendaya et François Hollande ils doivent rien avoir à se raconter quand ils se retrouvent en repas de famille.

[Solvaix @Solvaix]

3 HOURS OF MANUAL SETUP GOT AN AI BUILDING INSIDE REVIT. AUTODESK'S OWN VERSION STILL CAN'T.

Revit is the software behind most MEP work, the mechanical, electrical, and plumbing systems planned out digitally before a single pipe goes into a real building. Autodesk released Revit 2027 in April 2026, then followed it in June with an official MCP server, Model Context Protocol, letting outside AI tools like Claude plug directly into a live model instead of guessing from screenshots.

Claude Fable 5 connected without issue. It could read walls, ducts, pump schedules, dozens of parameters per object. It couldn't create a single wall, duct, or schedule, Tech Preview status, write access still locked on purpose.

A beginner MEP engineer wanted more: three pumps, a chiller connection, basic ductwork, built by description instead of menu clicks. The official server couldn't do it.

The fix was a community-built Revit MCP project instead, unofficial, 61 separate functions in total, including the ability to actually create and edit elements, the access Autodesk hasn't opened up yet. No official installer, no drag-and-drop, real technical setup before anything worked, close to 3 hours in this case.

Once it was live, a small MEP layout that would've taken most of a workday to build by hand was finished in 40 minutes.

Over in 3D art and design software, that gap doesn't exist. Blender's official MCP has had full read and write access since April 2026, backed by Anthropic as a funding partner of the Blender Foundation. Get a 3D scene wrong and you hit render again. Get a real pipe run wrong and you're filing a change order. That's the caution Autodesk is building into Revit.

[David @daavidaviid]

@cleeeeeeeeement J'utilise Argent c'est pas mal.
https://t.co/FSkaCBUDN3

[Rand @rand_longevity]

AI is not just coming for your job, it is coming for the entire damn system

the system of going to high school, college, degrees, résumés, and the 40-hour week is now coming to an end

[Fav @Favwontmiss]

i learned ants really have no concept of us as humans because of their limited eyesight and how big we are to them. they see us as moving shadows, or wind, or like obstacles and landscape (i imagine mountains). makes me wonder what’s in front of me right now that’s too large for me to comprehend.

[Kahris @chrissotraidis]

Ocarina of Time, running natively on iOS and iPadOS for the first time.

Not an emulator. Ship of Harkinian has run on Windows, Linux, macOS, Switch, Wii U, and Android. Never iOS. Now it runs on iPad, with iPhone in the build too.

The game was decompiled to C and rebuilt as a native ARM64 app rendering through Metal. No N64 hardware being simulated, no ROM interpreted at runtime, no streaming from a Mac.

It isn't stuck at the N64's 240p and 20fps. Widescreen, full resolution, 60fps.

This game deserves a controller, so controller support is native. Pair a Bluetooth controller and play the whole thing. Touch, keyboard, and mouse work too.

I used Codex 5.6 Sol to build the iOS layer. Open sourcing it all below.

[naiive @naiivememe]

eggs watching the rest of the fridge from their private balcony

[Ludonauta @LudensLudonauta]

I was studying MCPs and this comment really blew my mind.

[Polymarket @Polymarket]

JUST IN: Math PhD student solves 6 open Erdős math problems in 5 days using GPT-5.6 Sol.

[Uncle Bob Martin @unclebobmartin]

I’m significantly older than you. I started coding in the late 60s. My current strategy is to not read any of the code written by my agents. That’s the only way I can take advantage of their productivity. What I do instead is to surround the agents with extreme constraints. Unit tests, gherkin tests, QA procedures, quality metrics, mutation testing, test coverage, and a plethora of others. In the end, I have very high confidence in the code they produce because they’ve had to run the gauntlet of all of my constraints and tests.

[David Haz @davidhdev]

Introducing Canvas UI, the first ever html-in-canvas component library.

Your DOM is the render target now. Real-time shaders over real, interactive UI.

24 components. React, Vue, Svelte, vanilla TS.

Free. Open source. 🧵
https://t.co/Y5ZZDNxPUX

[Xavier Niel @Xavier75]

Le hasard oui https://t.co/vOnczowxRT

[David @daavidaviid]

@reczko_konrad Awesome! Any resource/skill on how to do something similar?

[Guillermo Rauch @rauchg]

Everyone is now a programmer, because the present and future programming language is your natural language. If you can write or speak, you can build. This is the most important evolution in our field since its creation. And it's a tsunami that has *just* started.