Yes, you can build machine learning and even simple neural network models using a smartphone, but not by relying on the phone’s hardware itself. Instead, you should use cloud-based tools.

The best option is Google Colab, which allows you to run Python code (including libraries like NumPy, TensorFlow, and PyTorch) directly from your phone’s browser. It even provides free GPU access in some cases.

You can also use Hugging Face to access pretrained models and APIs, which is especially useful since training large models from scratch is not practical on limited hardware.

There are also apps like Pydroid 3 or Termux for local Python coding, but they are limited and not suitable for heavy ML workloads.

In practice, your phone acts as a client while the actual computation happens in the cloud, which is the most efficient way to learn and build ML models without a PC.