็ฎไฝไธญๆ
EnglishWhat is Made With ML by Anyscale?
Updated
Made With ML is a free educational platform for developers who want to learn how to design, develop, deploy, and iterate on production machine learning applications. It combines machine learning concepts with software engineering best practices, and is created by GokuMohandas and associated with Anyscale. The site states it has 40K+ developers learning on the platform. It's best suited for developers who already have some Python and ML familiarity and want to move from notebooks to production-grade systems.
What the platform covers
The curriculum is organized into eight sequential modules:
| # | Module | Topics |
|---|---|---|
| 1 | ๐จ Design | Setup, Product, Systems |
| 2 | ๐ข Data | Preparation, Exploration, Preprocessing, Distributed |
| 3 | ๐ค Model | Training, Tracking, Tuning, Evaluation, Serving |
| 4 | ๐ป Develop | Scripting, Command-line |
| 5 | ๐ฆ Utilities | Logging, Documentation, Styling, Pre-commit |
| 6 | ๐งช Test | Code, Data, Models |
| 7 | โป๏ธ Reproducibility | Versioning |
| 8 | ๐ Production | Jobs & Services, CI/CD workflows, Monitoring, Data engineering |
There are also Foundations lessons covering Toolkit (Notebooks, Python, NumPy, Pandas, PyTorch) and Machine Learning (linear regression, logistic regression, neural networks, data quality), plus Deep Learning (CNNs, Embeddings, RNNs, Attention, Transformers).
How it teaches
The course follows a stated set of principles:
- First principles โ every ML concept is developed from the ground up before jumping into code.
- Best practices โ software engineering practices are applied throughout model development and deployment.
- Scale โ ML workloads (data, train, tune, serve) scale in Python without learning entirely new languages.
- MLOps โ components like tracking, testing, serving, and orchestration are connected into an end-to-end system.
The specific task used in the course is fine-tuning an LLM for a supervised task, but the site notes the lessons extend to other applications (NLP, CV, time-series), models (regression through LLMs), data modalities (tabular, text), cloud platforms (AWS, GCP), and scale (local laptop to distributed cluster).
Who it's for
The content is aimed at developers. If you're comfortable with Python and have some ML exposure but want to understand how to responsibly deliver value with ML in production, this is the target audience. Complete beginners may want to work through the Foundations section first.
Access and cost
The site offers a Subscribe option and mentions a live cohort where participants get live lessons, QA, compute (GPUs), and community. The homepage does not list pricing for the cohort or specify what the subscription includes, so check the site directly for current terms. The lesson content itself is presented as viewable on the site.
Where to start
- Go to madewithml.com and review the course lesson list.
- If you're new to the tooling, start with Foundations (Toolkit โ Machine Learning โ Deep Learning).
- Otherwise begin at Module 1 (Design) and work through the modules in order.
- The code repository is referenced as
GokuMohandas/Made-With-MLfor following along.
The main thing to decide up front: whether you want the self-paced lesson path or the live cohort with GPUs and instructor QA. The former costs nothing to browse; the latter requires signing up and checking current availability.