Ever felt like this?

There is (very) intelligent life beyond LLMs...

Without data engineering...

... there is no LLM engineering.

Today is the first day of the rest of your life.
Choose your path wisely.

How do you monitor your ML models?

Rate my setup

→ Inference pipelines, that load the model from the registry, and the input from the client app (for example a vector of numerical features, or a text prompt), generate a prediction (or a generation) and return it to the client app.

→ Training (or fine-tuning) pipelines, that read historical features from the Feature Store/Vector DB and generate a new model artifact, either by training from scratch or fine-tuning a base LLM. This model artifact is then pushed to the model registry.

Any ML system can be decomposed into at least 3 types of programs (aka pipelines)

→ Feature pipelines, that transforms raw data into ML model features (e.g. vector embeddings) that are saved in a Feature Store or Vector DB.

Don't now how to start an ML project?

This is the way ⬇️

4 years ago I struggled to land my first freelance ML engineering contract.

Then I discovered this ↓

Deploying ML models is 𝗲𝗮𝘀𝘆...
... when you use the 𝗿𝗶𝗴𝗵𝘁 𝘁𝗼𝗼𝗹𝘀

You choose

Advice for ML beginners 💡

→ A machine learning model is a product of the dataset it was trained on.

→ Bad data means a bad model.

→ Good model means a good dataset.

The First Rule of Machine Learning ⬇️

Claude aged really well