Newsletter Blog Career Sessions
← Back to all posts

#15: Deep Learning and Transformers - the roadmap | Visibility > technical skills - why?

by Timur Bikmukhametov
May 23, 2025

 

Reading time - 7 mins

 

🥇 Picks of the Week

New best LLM for coding, baseline library for anomaly detection, and more.

 

🧠 ML Section

Deep Learning & Transformers - the roadmap

 

💰 ML Section

Why visibility > technical skills in getting ML jobs


Land your next ML job. Fast.

​I built the ML Job Landing Kit to help Data Professionals land jobs faster!

✅ Here’s what’s inside:

- 100+ ML Interview Q & A

- 25-page CV Crafting Guide for ML

- 10-page LinkedIn Guide for DS/ML

- 2 Ready-to-use CV Templates

- $40 discount for a TOP Interview Prep Platform

This is the exact system I used to help 100+ clients land interviews and secure ML roles.

🔥 ​Grab ​​your copy​​ & land your ML job faster!

 


1. ML Picks of the Week

⏱ All under 60 seconds. Let’s go 👇

🥇 ML Tool

→ How to quickly run an Anomaly Detection baseline? Use PyOD, it has 45 anomaly detection algorithms!

 

📈 ML Concept

→ Stationary vs Non-stationary Time Series: Differences, Similarities, Examples 

 

🤔 ML Interview Question:
→ Which classification metrics are best to use for imbalanced datasets?

One-line answer: Use F1 score - it balances precision and recall, giving a better picture than accuracy.

Learn more here about imbalanced classification problems.

 

🗞️ One AI News

Anthropic released Claude 4 - the new best model for coding. 

 

🧠 Weekly Quiz: 

→ Which condition is NOT listed in linear regression assumptions?

A) Normal distribution of features

B) Independent residuals

C) Constant learning rate

D) Homoscedasticity of errors

​✅ See the correct answer here

 


2. Technical ML Section

Deep Learning & Transformers - the roadmap

People often try to learn deep learning by copying code from Hugging Face or running GPT-like models without understanding what’s going on inside. 

That might work for toy projects, but in real-world systems, this lack of intuition becomes a bottleneck.

This is a 5-step roadmap that I recommend if you want to actually understand how deep learning and transformers work, and not just use them like a black box.


📌 Step 1: Learn Core Deep Learning Concepts

You need a solid foundation. That means you must understand how neural networks are trained, what backpropagation does, and how architectures like CNNs and RNNs process data.

Andrew Ng’s Deep Learning Specialization on Coursera is still one of the best resources for that.

The explanations are clear, math is digestible, and there’s a strong focus on intuition.

👉 Deep Learning Specialization (Coursera)


📌 Step 2: Internalize with Code

This is where you build intuition with code.

The fast.ai course flips the typical approach—you start coding real models right away using PyTorch, and the theory is introduced as needed.

It’s super practical and designed for people who want to build, not just study. You’ll train image classifiers, NLP models, and tabular networks from Day 1.

👉 fast.ai Deep Learning for Coders


📌 Step 3: Master PyTorch

If you’re building DL models, you’ll likely use PyTorch. 

The syntax is clean, debugging is easy, and it's the standard in most research and production codebases today.

The official PyTorch tutorials are underrated.

They don’t just teach API usage, but they walk you through the internals of model building, training loops, gradient tracking, and module design.

You’ll write your own custom layers and learn to debug weird training issues, which is crucial for real-world work.

👉 Official PyTorch Tutorials


📌 Step 4: Understand Transformers Visually

Jay Alammar breaks down the transformer architecture step by step: token embeddings, positional encodings, attention heads, and more.

Once you go through his “Illustrated Transformer”, you stop just running BERT and start understanding what it’s doing.

👉 The Illustrated Transformer by Jay Alammar


📌 Step 5: Real-World Transformers

The Hugging Face course and docs are the most practical resource for applying transformers today.

You’ll learn how to fine-tune pre-trained models like BERT, GPT, and T5 on your own datasets.

It covers tokenization, training loops, evaluation metrics, and even deployment basics.

This is exactly the kind of work companies look for in NLP projects—and the stuff that gets you interview-ready too.

👉 Hugging Face Transformers Course


That's it for the Technical Part! 

Follow me on LinkedIn​ for more daily ML breakdowns.


2. ML Career Section

Why visibility > technical skills in getting ML jobs

Most ML folks are obsessed with one thing: becoming technically better.

Sure, learning new loss functions, reading another paper, tweaking another pipeline… all that matters.

But here’s the hard truth: these days, jobs don’t go to the best technical people. They go to the most visible.

If your name shows up a few times on LinkedIn - shared a post about your project or wrote a quick ML breakdown- chances are, a hiring manager or recruiter has already noticed you.

And that alone puts you ahead of 99% of candidates who stay invisible.


So what does visibility look like?

There are many possibilities. Here are a few of them:

  • Posting a weekly LinkedIn breakdown of what you’re learning (even simple stuff)

  • Sharing short write-ups of past projects.

  • Creating a public roadmap of your ML learning and sticking to it.

  • Publishing a practical tutorial that you wish had existed when you were learning

You don’t need to go viral. You need to be findable.

Because in today’s job market, being visible is what gets you into the room.

Your skills matter after that.

What to know how you can become more visible?

Book a 1:1 Career Session with me and we'll figure out a personalized path for you.


That is it for this week!

If you haven’t yet, follow me on LinkedIn where I share Technical and Career ML content every day!


Whenever you're ready, there are 3 ways I can help you:

​

1. ML Job Landing Kit

Get everything I learned about landing ML jobs after reviewing 1000+ ML CVs, conducting 100+ interviews & hiring 25 Data Scientists. The exact system I used to help 70+ clients get more interviews and land their ML jobs.

 

2. ML Career 1:1 Session 

I’ll address your personal request & create a strategic plan with the next steps to grow your ML career.

 

​3. Full CV & LinkedIn Upgrade (all done for you)​

I review your experience, clarify all the details, and create:
- Upgraded ready-to-use CV (Doc format)
- Optimized LinkedIn Profile (About, Headline, Banner, and Experience Sections)


Join Maistermind for 1 weekly piece with 2 ML guides:


1. Technical ML tutorial or skill learning guide
2. Tips list to grow ML career, LinkedIn, income

 

Join here! 

 

 
 

 

 

 

 

 

#14: Feature Selection with SHAP - Intro | TOP 3 CV Mistakes
Reading time - 7 mins   🥇 Picks of the Week Library for torch, Data Drift Importance, ML Interview & ML Quiz Questions, and more.   🧠 ML Section Begin using SHAP values for feature selection (with code example).   💰 ML Section Learn the TOP 3 CV Mistakes and how to avoid them Together with datainterview.com​ Become 100% ML Interview Ready. Now. ​DataInterview.com​ helps you finally nail ML ...
#13: 3 Simple Steps to Filter Outlies | TOP 3 ML Portfolio Mistakes
Reading time - 7 mins   1. ML Picks of the Week ML Tool, Concept, Resource Link, Post, Interview Question, and Quiz of the week. All in under 1 minute.   2. Technical ML Section Learn 3 Simple Steps to Filter Outliers   3. Career ML Section Learn TOP 3 ML Portfolio Mistakes 1. ML Picks of the Week ⏱ All under 60 seconds. Let’s go 👇 🥇 ML Tool → Pyro - Python probabilistic library built on PyT...
#12: 5 Cases When Gradient Boosting Fails | Why NOT to be an ML/DS Generalist
Reading time - 7 mins   1. ML Picks of the Week A weekly dose of ML tools, concepts & interview prep. All in under 1 minute.   2. Technical ML Section Learn When Gradient Boosting Fails for Tabular Data   3. Career ML Section Learn why NOT to be an ML/DS generalist to make a successful ML career 1. ML Picks of the Week 🥇ML Library tsfresh Tsfresh is a powerful Python library for automated fe...
Powered by Kajabi