Had a blast recording this podcast with @farisaziz.com , Staff Eng at Smallpdf and organizer of @zurichjs.com . We talked about engineering fundamentals, the story behind ZurichJS, career lessons, what moved the needle for him & what turned out to be a time waster. youtu.be/xDCBvtUvLtg

Faris Aziz – Staff Engineer at @Smallpdf & @zurichjs organizer

YouTube video by Js-Craft Academy

youtu.be

My book "📙 Understanding Neural Network - From Writing if else Code to Training AI Models" is 50% done. I'm currently reviewing the draft of the first part. Remember that you can join the waiting list and get an early-bird 20% discount 👉 www.js-craft.io/nn/

Token - maximizing ... OpenAI gives awards to the companies that use the most tokens. That's kind of like McDonald's rewarding you for buying the most Happy Meals, or Ford rewarding you for buying the most cars. Why is "using more tokens" is a good thing?

Getting ready to record a podcast later today with @farisaziz.com , organiser of Zurichjs and Staff Engineer at SmallPDF. Drop your questions below, and I’ll ask Faris as many as I can.

"I haven't written code in months" is an odd thing to brag about if you still call yourself a software engineer. Don't get me wrong! I use AI every day, and it's incredible. But AI helps you move faster; it doesn't eliminate the need to stay close to the code, understand the system, and stay sharp.

This is the best AI-related talks I’ve heard in months. Middle grounded discussion, based on a lot of real data. I’m tired of armies of techno optimists selling superabundance (plus AI stocks) or doomers forecasting Mad Max without any real logic. www.youtube.com/watch?v=BD3v...

The most rational take on AI you’ll hear this year

YouTube video by Lenny's Podcast

youtube.com

I'm delighted to share my interview with @joelhooks.com, co-founder of Egghead and one of the most down to earth voices in developer education space. Topics we cover: – learning in the AI era – validating courses & products – business, bets & Egghead – long-term projects youtu.be/MLAJy8-xrUM

Joel Hooks — Egghead.io &ex Head of Education @Vercel, dev education and staying true to your values

YouTube video by Js-Craft Academy

youtu.be

Gradient descent is like finding the bottom of a valley blindfolded. At each step, you can feel just the current slope: if slope < 0 → increase weight if slope > 0 → decrease weight if slope == 0 → you're at the min error 👉 www.js-craft.io/nn

The best morning are the ones when I can wake up before the noise of the day starts, make a cup of tea, and be able to do some deep work on something I enjoy a lot. Just pen and paper, no distractions for 90 min 👉 www.js-craft.io/nn/

Morning walk on the beautiful streets of Cordoba and a very inspirational listen with Faris Aziz at the Senors at Scale podcast. Loved the idea of seeing developer educational events as mini start-ups + many knowledge gems on the FE architectural side. Thank you Spain!

Doing the main research for my new book on Neural Networks for JS Developers entirely with pen and paper. No distractions, just deep work & black tea. Loving the process! The faster tech moves, the more strong fundamentals are becoming essentials 👉 www.js-craft.io/nn/

Training neural networks feels like playing 2 different games: -🎯 Guess the Correct Number, or trying to find the bottom of a valley while blindfolded (adjusting weights) -⚖️ The Blame Game: figuring out how much each neuron contributed to the prediction error (backpropagation) www.js-craft.io/nn/

Understanding Neural Networks Book Presale 20% off

JavaScript, LangChain and AI integration resources.

js-craft.io

Hrishi Olickel, former guest of the JS Craft Podcast, gave a great talk at the AI Engineer Singapore - timestamp 08:30:00 - on adding reliable agentic systems into the middle of the stack instead of the tip. Check it out! High-quality, no hype-driven info 🥇 www.youtube.com/watch?v=m12v...

AIE Singapore Day 2 ft. Google DeepMind, OpenClaw, Adaption, Arize, Cloudflare, Robot Company & more

YouTube video by AI Engineer

youtube.com

Great topics - especially with Chrome 148 adding the Prompt API.

Chris Coyier@chriscoyier.net · 3mo ago

My next ten thousand dollar idea is Chris ‘n’ Daves Website Salvage Yard. You give us old websites you can’t keep up anymore, we slather them in ads, but promise to keep them online in perpetuity.

Saying we don’t need to learn to write because LLMs can do it is like saying we don’t need to go to the gym because machines can lift for us. Writing is for the mind what exercise is for the body. Some good points here youtu.be/6pxmdmlJCG0

OpenAI Founder Breaks Down His Writing Process — Sam Altman

YouTube video by David Perell

youtu.be

I would add to this list also LLMs;
 LLM that doesn’t know it doesn’t know + confident human who also doesn’t know what they don’t know = very expensive mistakes Deep fundamental understating matters more that ever right 👉 www.js-craft.io/nn/

The books of Nassim Taleb need be studied as thoroughly as high school textbooks. Very few books have changed and deepened my view of the world as the Incerto series.

I think that most applications will soon include a built-in AI/LLM layer. Like they have a database layer today. The build in LLM could be used for: - 💬 natural language UI interactions - 📦 working with unstructured data - 🤖 communication with other LLM-powered systems

The neuronal network model paper was written in 1943. Still relevant. Still foundational. Still powering modern AI. That’s Lindy knowledge worth investing in. I'm working on a book for devs who want to understand AI from the ground up by building neural networks from scratch js-craft.io/nn

Understanding Neural Networks Book Presale 20% off

JavaScript, LangChain and AI integration resources.

js-craft.io

I’m seeing more and more GitHub files that mix natural language with code. It’s making me think if Andrej Karpathy may be right; maybe the next world’s most popular programming language could be English.

OpenClaw interest over time. That's why I try to avoid investing time & effort into learning hype based tech and focus on the fundamentals that don't change 👉 js-craft.io/nn

LLMs are just machine learning models. It’s dangerous to combine an LLM that doesn’t know it doesn’t know with a human who also doesn’t know what they don’t know. That’s how you get confident, expensive mistakes. Real competence & understating matters more that ever!

Trends come and go. But the foundational Artificial Neuron Model paper was published in 1943. Still relevant today. The Lindy effect at work, knowledge this old is here to stay. And that's why I'm making a book for JS devs who want to truly understand AI 👉 js-craft.io/nn

Understanding Neural Networks Book Presale 20% off

JavaScript, LangChain and AI integration resources.

js-craft.io

Chrome 148 now has the Prompt API. You can now call a local Gemini Nano with .prompt('Write me a poem about cats'). Learned it from Stefan’s Web Weekly newsletter!

Stefan Judis@stefanjudis.com · 3mo ago

Web Weekly 191 will go out in just a little bit! This one covers `sizes=auto`, the `No-Vary-Search` header, `RegExp.escape()`, HTML in canvas, and new Git version that makes rewording commits easier. 🎉 webweekly.email/archive/web-...

Still got Private Pens and they work the same way, but now there is 📝 password protected Pens and 🎟️ invite-only Pens.

CodePen@codepen.io · 3mo ago

We've long had Private Pens on CodePen, but we've taken it 2 notches up.

My new book project - 📙 Understanding Neuronal Networks from Scratch - has now a landing page! Learn more about the book and get 20% off at launch 👉 www.js-craft.io/nn/

Understanding Neural Networks Book Presale 20% off

JavaScript, LangChain and AI integration resources.

js-craft.io

Daniel - Js Craft@danielnsts.bsky.social · 3mo ago

Something new is cooking ✍️

Coworking view! Thank you, Spain!

Something new is cooking ✍️

is training AI models the new coding ?

Daniel - Js Craft@danielnsts.bsky.social · 3mo ago

Training a neural network sometimes feels like designing a giant Pachinko machine. You build the structure, pour in data, and watch where the balls land. It's programming just not the kind I was thought in school. youtube.com/shorts/aygM1...

Training a neural network sometimes feels like designing a giant Pachinko machine. You build the structure, pour in data, and watch where the balls land. It's programming just not the kind I was thought in school. youtube.com/shorts/aygM1...

Vintage Pachinko machine – with monkey bird and angry hippopotamus, it works

YouTube video by isbjorn25

youtube.com