No BS AI/ML Content | ML Engineer with a Plot Twist 🥷30M+ Views 📝

Done-for you Audience Building & Content Writing for AI/ML Founders that frees up 100% of your time, so you can focus entirely on building your company  
  
Is your work too important to ignore? 💌 https://tally.so/r/wk69VJ  (or shoot me a DM) 👉
  
It's 2025 and we're at the peak of the attention economy  
  
Where Authenticity is king, visibility is currency and people wanna buy from people

Not faceless dehumanized brands  
  
So, they told you LinkedIn was "free marketing" you slapped some posts and called it strategy  
  
You've poured hours crafting the 'perfect' post, only to watch it sink without a trace..  
  
Then it dawned on you, your technical brilliance isn't enough to draw a crowd  
  
Hell, maybe you've even tried mimicking influencers, scrambling for any traction..  
  
Your post got 500 f*cking views... from your friends...  
  
Being ignored is soul-crushing  
  
It's not your fault tough  
  
When you started your company, you didn't sign-up for a degree in an Audience Building, Copywriting and LinkedIn strategy 

And here you are, sacrificing yet another late night for your vision  

"Am I just sabotaging my efforts?"  
  
"Is my life-work meaningless?"  
  
Getting traction feels impossible  
  
You hoped hiring some marketer would save your weekends but..  
  
Tech is a code they can't crack  
  
They can only churn out the lame, soul-sucking crap we all ignore on sight  
  
With armies of founders flooding feeds, your voice dissolves into meaningless white noise  
  
But in the process, you've wasted more of your precious time explaining 101 tech concept and rewriting low-quality drafts 

Done being ignored?  
  
Let's make the world listen 
  
I can help you with:  

✒️ No BS AI/ML Content  
⭐ LinkedIn Profile Revamp
🎯 Website Copy Optimization  

I've done it before:

↳ For founders backed by YC, a16z, Sequoia, and Techstars
↳ For bootstrappers in LLMs, ML, Vector DBs, Data Analytics, and Engineering
↳ For AI and Machine Learning leaders at Google, Microsoft, Airbnb, and YouTube

If they could rise above the noise, so can you

Yeah, even if you're an introvert with a "boring" niche

Still with me?  
  
Amazing—You're amazing!  
  
Cheers,  
Paolo

My booking link: 💌 https://tally.so/r/wLW0oz (or shoot me a DM) 👉

Paolo Perrone

They told you 70B models need 80GB VRAM. They lied.

AirLLM runs Llama 70B on a 4GB GPU.

No quantization. No distillation. No pruning.

The trick: layer-wise inference. Instead of loading the entire model into VRAM, it loads one layer at a time from disk.

Your GPU only ever holds one layer. The rest lives on disk.

What you can run:

→ Llama 70B on 4GB VRAM
→ Llama 405B on 8GB VRAM
→ Qwen2.5, Mixtral, ChatGLM all supported

pip install airllm

from airllm import AutoModel

model = AutoModel.from_pretrained("garage-bAInd/Platypus2-70B-instruct")

input_tokens = model.tokenizer(input_text,
    return_tensors="pt",
    truncation=True,
    max_length=128)

output = model.generate(
    input_tokens['input_ids'].cuda(),
    max_new_tokens=20)

Standard transformers API. That's it.

Optional: 4-bit compression for 3x speed boost.

model = AutoModel.from_pretrained("...", compression='4bit')

Trade-off: slower inference than full VRAM loading. But if you don't have the VRAM, you don't have options.

Works on MacOS with Apple Silicon too.

1.9k GitHub stars. Apache 2.0.

→ github.com/lyogavin/airllm

💾 Save for when your GPU budget doesn't match your model ambitions
♻️ Repost if you've ever closed a tab because "requires 80GB VRAM"

No BS AI/ML Content | ML Engineer with a Plot Twist 🥷100M+ Views 📝

Failing my way to success | Data Entrepreneur

Meet Shaw, an AI educator, data scientist, entrepreneur, researcher, drummer, and bread enthusiast driven by an insatiable curiosity to understand the world and create better systems.

With a Ph.D. in Physics from The University of Texas at Dallas, Shaw’s research focused on novel AI applications to fields spanning neuroscience, environmental science, and human performance.

As an educator, Shaw is driven by a passion for learning and a mission to make AI accessible to all. His YouTube channel and technical blog have taught millions and have gained an audience of over 100,000 builders.

Shawhin Talebi

15 months ago, I launched the AI Builders Bootcamp.

Today, I'm closing it down.

Software development has changed. 

Before launching an app or SaaS meant months of work and assembling the right team. Now, one person can ship the same thing in a weekend using AI.

This lower cost of building allows everyone to think bigger.

- The non-coder can spin up prototypes without hiring a dev
- The coder can build a full-stack app in weeks
- The experienced developer can ship 10X faster

While the AI Builders Bootcamp has been the most rewarding experience of my entrepreneurial career, I also need to think bigger.

That's why the cohort starting this Friday (Jan 16) will be my last one ever.

I'll share more about what I'm building in the coming weeks...

But if you're seeking the confidence and clarity to build anything you can imagine using AI, then join us this Friday!

👉 Secure your spot + start shipping in 2026: https://lnkd.in/gPHZTdWs

Shaw Talebi

How to ship your AI app in 2026 🚀

Although building software is easier than ever, most AI projects still stall halfway through. This is because there's more to building an AI app than just describing it to Claude.
​
Here's the 5-step process I use to ship ideas without getting stuck.
​
Step 1: Have an idea
​
While this step might seem obvious, there is a common mistake people make here. Namely, they don't solve a real problem.
​
This is important because value is generated, not by creating software, but by solving problems. And if you're going through the trouble of building something, you might as well make it valuable to someone (even if it's just you).
​​
Step 2: Scope the MVP
​
Another common mistake people make is trying to build too much too early. Then, halfway through, they realize the project is much more work than they thought and abandon it.
​
Building AI apps is an iterative process. That's why I always start with a minimal (yet complete) version that I can implement in a pre-defined timebox e.g. 2 hours.
​
This forces me to focus on the app's essential features and build them first.
​
Step 3: List key steps
​
Once I've decided what I want the MVP to do (and not do), I then break it down into key steps. Usually, this is some version of the following:
​
>> User provides input → App does something → App produces output
​
Mapping this out forces me to think through how I actually want the app to work under the hood (and gives more clarity on how to build the thing).
​
Step 4: Pick tech stack
​
There are many ways to implement any given step in a workflow. That's why my next step is to narrow the scope a bit more by picking the tech stack I want to use.
​
My rule of thumb is to build with what I know, so Python + any library I've used in the past. If I don't know of any libraries that seem helpful, I'll research tried-and-true ones.
​
Pro Tip: The number of GitHub repo stars is a good eval metric for unfamiliar libs
​
Step 5: Build it!
​
Now that I have a good understanding of what I want to build and how I want to build it, it's time to execute.
​
My first step is to talk through the previous steps with ChatGPT (the best dictation among AI tools) and have it write a spec sheet for the project. After refining the spec, I'll paste it into my project folder and use Claude Code to start building.


What's something you want to build in 2026?

In 2024, Anthropic released the Model Context Protocol (MCP) to help connect AI models to the right tools and context.

Now they have "Agent Skills" to help models use these tools more effectively (among other things).

Agent Skills = reusable instructions an agent can access as needed

It's key feature is so-called "Progressive Disclosure" i.e. information is given to the AI just-in-time.

This resolves one of the big issues with MCP which was bloating the context window with (potentially irrelevant) information at the start of every conversation.

In recent a YouTube video, I talk more about how Skills work and how to use them in Claude Code.

Check it out: https://lnkd.in/geewh_ZD

I was talking to a CXO at a large org about turning the promise of AI into profit.

And they said something that really stood out to me…

There is a major disconnect between their personal experience with AI and rolling it out to their teams.

For example, they might spend 1 hour spinning up a website with Claude Code, but when trying to scale it across all their teams, there is no clear impact on business KPIs.

This is just one of many challenges I’ve heard from leaders in recent chats.

But I’m curious…

Is this something you’re seeing too? Why do you think it’s happening?

5 Mistakes AI Builders make (and how to avoid them)

A lot has changed in AI since I started 8 years ago.

However, there is (surprisingly) much that has stayed the same. This includes the mistakes people make when it comes to building AI solutions.

Here are the ones I see come up over and over again.

Mistake 1: Start with tech, not problems

If value is the goal, then starting point for the project shouldn't be technology, but rather a problem worth solving.

Mistake 2: Over-complicating v0

Your first version should be simple enough to let you iterate and run experiments quickly.

Mistake 3: Over/Under Planning

Too much planning and you never ship. None at all and what you ship sucks.

Talk to users, define the problem, scope the solution, and set milestones. But don't try to figure out all the details before building.

Mistake 4: Ignore Evals

Without evals, you don't know if your solution is any good. Measure so you know what to fix.

Mistake 5: Prioritizing Accuracy Over Impact

A high performance score doesn’t always mean your app is valuable for users. Real evaluations should be grounded in impact not accuracy. 
 

In a recent YouTube video, I talk more about these mistakes and how to avoid them.

👉 Check it out here: https://lnkd.in/gW5HWV8h

Skills and MCP give us ways to customize agents with tools and context.

While these technologies overlap, they also have key differences and distinct use cases.

Here’s the short version:
- MCP = gives tools to agents
- Skills = teaches agents how to use said tools (among other things)

In a recent YouTube video, I discussed these differences in more detail.

👉 Check it out here: https://lnkd.in/gy2PQgYt

While it's never been easier to build software, this has created a new problem...

... Vibe Coding Hell 😈

In other words, generating code faster than you can understand (and debug) it.

But it doesn't need to be that way.

This Friday (Jan 9) at 10AM CST, I'm hosting a live lecture on how to transform Claude Code from a coding agent into a Python tutor.

You'll learn how to customize Claude with output styles and come away with a reusable tutor template you can apply to other learning contexts.

👉 Join me live and ask me your AI questions: https://lnkd.in/gnfq9-TJ

I’ve been chatting with leaders about how AI is showing up in their orgs.

One of the major themes from these conversations is that the profits from AI have not lived up to the promise.

Most companies have already implemented AI.

Whether that’s...
- Building an internal chatbot
- Rolling out Microsoft Co-pilot to team members
- Giving developers access to Claude Code, Cursor, or the like
- Or even deploying a local LLM inference stack

However, there’s often little to no impact on key business metrics.

I’ve been thinking a lot about this and finally came to an obvious realization.

10X impact from AI won’t come from just adding the technology to existing workflows, but rather rebuilding workflows from 1st principles using LLMs.

Doing the former is like buying computers for all your employees, but they still do all their work with pen, paper, and filing cabinets.

It wasn’t until the work became digitized that computers made an impact. 
And it’s the same case for AI.

I’m curious how other people are thinking about this.

What do you see as the gap between current AI implementations and real-world impact?

taplio

The Battle

Paolo Perrone

Shawhin Talebi

Followers

Posts

Engagements

Likes

Performance ScorePaolo Perrone's LinkedIn Performance

Performance ScoreShawhin Talebi's LinkedIn Performance

Ready to grow your LinkedIn presence?

taplio

The Battle

Paolo Perrone

Shawhin Talebi

Followers

Posts

Engagements

Likes

Performance Score.css-goub6s{font-size:var(--chakra-fontSizes-md);font-weight:500;color:var(--chakra-colors-gray-600);display:block;margin-top:var(--chakra-space-1);}Paolo Perrone's LinkedIn Performance

Performance ScoreShawhin Talebi's LinkedIn Performance

Ready to grow your LinkedIn presence?

Performance ScorePaolo Perrone's LinkedIn Performance