Kemal Cholovich

📈 Contractor #data #ai #ml #genAI
☁️ Founder @GdgSarajevo

[Andrew Ng @AndrewYNg]

The most important skills in Building and Deploying AI Applications. https://t.co/IyWIKLIzeM

[FDE Jobs @fdejobsX]

The US leads #FDE demand, but the demand is global.

#hiring #palantir #softwareengineer

[It's FOSS @Itsfoss]

Remember to give back to the open-source projects you rely on. 🐧❤️

[MIT CSAIL @MIT_CSAIL]

All major neural networks, in one chart.

v/The Asimov Institute

[The Economist @TheEconomist]

Artificial intelligence has turned everyone into a creator. From music to literature, we tracked how the technology is changing different fields. Discover the full story. Register to read for free https://t.co/W2xaRlLR5Y

[Vala Afshar @ValaAfshar]

Tom Brady: To be successful at anything, the truth is, you do not have to be special.

You just have to be what most people aren not: consistent, determined and willing to work for it. No shortcuts.

[Vala Afshar @ValaAfshar]

A masterclass on ‘how to speak’ by MIT professor Patrick Winston - 15 key lessons

[Vala Afshar @ValaAfshar]

A masterclass on conflict resolution by Steve Jobs

[vladuah @vladuah]

In 1997, a developer publicly told Steve Jobs he had no idea what he was talking about. Jobs paused, then answered a question nobody had actually asked

The developer's question was an insult wrapped in a technical challenge - asking Jobs to explain Java against OpenDoc, then adding: "Perhaps you could tell us what you personally have been doing for the last 7 years."

Jobs didn't defend the technology.

He didn't attack the developer either.

He agreed with him first.

"People like this gentleman are right in some areas," Jobs said. He admitted OpenDoc probably did things nothing else could do. He admitted the demos would look impressive.

Then he pivoted to the actual question underneath the insult: how does any individual piece of impressive technology fit into a vision large enough to sell $8,000,000,000 to $10,000,000,000 of product a year?

Here's the framework he laid out live, under pressure, with no preparation.

Start with the customer experience. Work backward to the technology. Not the reverse.

"You can't start with the technology and try to figure out where you're going to sell it."

He admitted he'd made that exact mistake - building from technology outward - more than anyone else in the room. "I've got the scar tissue to prove it."

Then he made the abstract point concrete with a story from over a decade earlier.

Apple built the first small laser printer in the US. Inside the box: the first Canon laser engine on American soil, custom controller software, Adobe PostScript, AppleTalk networking. Genuinely groundbreaking engineering.

None of that was the pitch.

Jobs held up the printed page and said: "Do you want this?"

Nobody needed to understand PostScript or AppleTalk to answer that question. In 1984, before laser printing existed for consumers, the answer was an immediate, involuntary "yes."

That's the whole model. The technology enables the outcome. The outcome is what you sell.

Jobs closed by conceding something most executives never say in public: "I readily admit there are many things in life that I don't have the faintest idea what I'm talking about."

Then he defended his team anyway - engineers being offered 3 times their salary elsewhere during the hottest hiring market in Silicon Valley, and staying.

"Mistakes will be made along the way. That's good, because at least some decisions are being made."

He didn't win the argument by being right about everything.

He won it by being right about the 1 thing that mattered: start with what the customer wants, then build backward.

[Andrew Ng @AndrewYNg]

New: A map of the most important skills in AI Engineering. https://t.co/VVkn1Dqp1N

[Real Python @realpython]

🐍📩 Modern OOP, bisect, Django Async, and More at Pycoders Weekly https://t.co/qobdGKnKro

[FlyRank @flyrank_ai]

Bing Webmaster Tools just dropped AI Performance, showing which pages get cited by Copilot & ChatGPT. Cool dashboard. But visibility without revenue is just a vanity metric. FlyRank turns those citations into a growth engine. DM me. 🚀 . .

[FlyRank @flyrank_ai]

ChatGPT is now an ad channel inside Shop Campaigns. High-intent buyers ask, see your product, purchase -one flow. No SERP, no tabs. Organic + paid AI presence is the new GTM stack. FlyRank owns the organic layer. The race is on. 🚀

[Codie Sanchez @Codie_Sanchez]

Seth Godin gave a masterclass on how to win in 2026 (without being lucky or miserable):

1. Before chasing any goal, ask yourself who taught you it was worth chasing. A surprising amount of what we call ambition is just trying to impress people we stopped caring about years ago..

2. Replace "but" with "and." You can build something great AND people will criticize it. You can take a risk AND fail. See those as part of the deal not reasons to stop..

3. Build a life you don’t need to escape from. Spending 50 weeks a year waiting for 2 of freedom is a terrible trade… fix the weekdays instead..

4. You don’t need millions of people. You only need to find a few thousand who'd genuinely miss you if you disappeared. That's more than enough to build an incredible business..

5. Trying to make everyone like your work is how you become average. The people who create something memorable almost always divide opinions..

6. Every rejection is feedback. Don't keep pitching harder. Listen to why people said no, then come back with something better..

7. Becoming a better leader starts with yourself. If you can't lead your own habits, your own decisions, and your own fears, nobody else will follow you for long..

8. Your past shouldn't decide your future. If you wouldn't choose it today, don't keep doing it just because you've already spent years on it..

9. Never quit during the hard part. Expect the hard part before you begin, push through it, then decide if it's still worth doing..

10. Most excuses sound reasonable until you say them out loud. Say it and you’ll find out whether they're real or just fear wearing a disguise..

11. Stop asking how many people saw your work. Ask how many people would've missed it if it disappeared. That's a much better measure of whether you're creating something valuable..

12. Some things are problems: a bad hire, low sales, losing a customer. Some are just situations: your age, the economy, or yesterday's decisions. Solve the first. Stop fighting the second..

I've admired Seth's work ethic for years so it was awesome to have him on the BigDeal pod. One of the few conversations I wanted to rewatch immediately.

Thank you @ThisIsSethsBlog

Search “BigDeal pod” on youtube to watch the whole episode.

[Financial Times @FT]

AI has opened up big holes in cyber security https://t.co/6imEU1bcc7

[Financial Times @FT]

Just published: front page of the Financial Times, UK edition, Friday August 14 https://t.co/mBPrtAnFwk

[Linas Beliūnas @linasbeliunas]

Instead of watching 1 hour of Netflix tonight, watch this ex-Google Chief Scientist Jeff Dean’s lecture. It’s the clearest explanation I’ve seen of the full AI engineering stack - from building LLMs from scratch all the way to one human coordinating 100 agents.

The best part is that it’s useful whether you’ve never touched a model or you’ve been shipping agent systems every day for the past year.

Bookmark it & watch the whole lecture this weekend, because it might end up being the most valuable thing you learn all week.

[Sebastian Raschka @rasbt]

Just saw that the LLMs-from-scratch repository passed 100,000 stars on GitHub!
This is super cool and motivating. I am really happy to see that this open-source repo has helped so many people.
Thanks also to everyone who shared ideas and opened PRs with improvements!

Of course, I plan to keep adding new material, including new attention variants and architectures (while bigger projects like RL and Reasoning From Scratch live in their separate repositories).

I am also currently working on a larger applied custom “small” LLM project. It has been keeping me super busy this month, but I will share more on that soon.

If you are new to it, some of the highlights include

1. Of course, the complete code path from tokenization and attention to pretraining, classification, and instruction fine-tuning, etc. All of it FROM SCRATCH, of course! (RL lives in a companion repo.)

2. From-scratch implementations of Llama, Qwen, Gemma, and Olmo (smaller variants that run locally and can be plugged into the training scripts).

3. From-scratch implementations of attention alternatives and other architecture components, such as GQA, MLA, sliding-window attention, Gated DeltaNet, DeepSeek Sparse Attention, cross-layer KV sharing, and mixture-of-experts

4. Materials on KV caching, training performance, memory-efficient weight loading, DPO, evaluation, and LoRA

So, if you don’t have any weekend plans yet, happy tinkering!

[Philip Kiely @philipkiely]

Inference Engineering launches today.

https://t.co/UNEIty1xZg

[Amir Haghighat @amiruci]

We just launched Baseten for Model Labs, a new platform designed specifically for labs building their own models. Let me explain why we did it:

People often ask us whether a closed or open model is best for them. The reality is that it depends on their specific use case: sometimes it's open, specialized/fine-tuned, or closed. All variants are, and will continue to be, essential parts of the AI stack.

That said, we can’t expect every model lab to become an enterprise-grade infrastructure company. Serving closed models at scale requires building the billing, security, compliance, authentication, authorization, compute procurement, regional support, and more. These are exactly the investments we’ve made at Baseten: first in Dedicated Inference, then in Model APIs and Training, and most recently in the Basten Frontier Gateway.

Baseten for Model Labs provides the infrastructure, compliance, and go-to-market support so that labs can serve and scale for the enterprise. It also provides enterprises with an easier way to use these models, without needing to onboard a new subprocessor or handle model licensing.

We believe that the future of AI will be an ecosystem where hundreds to thousands of models exist that solve different, specific problems. Our job at Baseten is to serve that future, and we’re proud to launch Baseten for Model Labs with over a dozen partners to help build a healthy multi-model ecosystem.