Is this article fiction? Does anyone else still have questions like this?
Vadim, you’d better just keep working quietly as a taxi driver.
Haven’t you broken your brain yet?
If you want to work in a government organization, you need a degree, and I think ours is better than none at all. You’ll move and enroll remotely in some German university.
If not, then follow the well-known roadmap and go for it. If you started your journey from the front line, you’ll work for 3-4 years and burn out, but what about the art of programming? Or how did you even get there?

What you need to know and be able to do as an ML Engineer
1. Programming

Python (basics: NumPy, Pandas, Matplotlib, Scikit-Learn)

Additionally: C++/Java (for production systems), SQL (for working with databases)

Skills in writing clean, optimized code

2. Mathematics

Linear algebra (matrices, vectors, operations on them)

Mathematical analysis (derivatives, gradients, optimization)

Probability theory and statistics (distributions, estimates, hypotheses)

3. Machine learning

Classic algorithms: regression, classification, trees, SVM, ensembles (XGBoost, Random Forest)

Deep learning: neural networks, CNN, RNN, transformers

Text processing (NLP), images, time series

Model quality metrics (accuracy, F1, ROC-AUC, etc.)

4. Infrastructure

Working with data: SQL, ETL processes

Docker, Kubernetes (containerization, orchestration)

Working in the cloud: AWS, GCP, Azure (for deploying ML models)

MLOps: CI/CD for ML, model monitoring, experiment management

5. Tools and libraries

TensorFlow, PyTorch (for neural networks)

Scikit-Learn (for classic ML)

MLflow, DVC (experiment management)

Airflow, Prefect (pipeline orchestration)

6. Soft Skills

Ability to work in a team (with Data Scientists, DevOps, analysts)

Solving applied business problems

Ability to document and explain solutions

Translated with DeepL.com (free version)