Advanced Project

This is the capstone phase of the course, where you turn everything you have learned into one working project. Instead of small exercises, you now choose a single, realistic goal and carry it from idea to a usable result. A good project follows clear phases: define the problem and success metric, gather and clean data, build and train (or configure) a model, evaluate it honestly, and finally deploy it so real users can interact with it.

You will pick ONE of three options. Option 1: build a custom chatbot that answers questions in a specific domain, often by connecting a large language model (LLM) to your own documents. Option 2: build an image recognition system that classifies or detects objects in pictures using a convolutional or pretrained vision model. Option 3: deploy an AI application to the cloud so it runs reliably and is reachable over the internet.

For example, an Ethiopian agriculture cooperative could build a chatbot that answers farmers' questions about coffee diseases in Amharic, grounded in the cooperative's own field guides. Or a clinic could build an image recognition system that flags possible skin conditions from a phone photo, then deploy it to AletCloud so health workers across regions can use it on cheap data plans.