Data Associate Programme
SMUBIA's flagship programme. Data-Associates build projects in small teams, and learn through teaching Machine Learning concepts to others.
The programme
Application based. Instead of sitting through lectures and tests, associates will learn through building real projects while being guided by mentors. At the same time, they will further their machine learning and data analytics knowledge through weekly Co-learning sessions, where they teach the curriculum to other associates.
AY 26/27 curriculum
Nine topics across one semester. No pre-requisite knowledge required. Together they map the essentials of modern machine learning.
Modelling continuous outcomes. Normal Equation, Loss, Regularisation, the basics of machine learning.
Drawing decision boundaries, from logistic regression to margin-based classifiers.
Bagging, boosting and forests. How weak learners can be combined into a strong one.
Collaborative filtering and matrix factorisation — the logic behind what you're shown next.
Backpropagation, Perceptrons and the building blocks that sit beneath modern deep learning.
Tokens, N-grams, embeddings and representing meaning as vectors.
Attention and transformers — the architecture under today's Large Language Models.
Convolutional architectures that allow a machine to read an image.
Agents that learn by acting: reward, policy and the exploration trade-off.
How it runs
Co-learning
The surest way to learn something is to teach it. Teams of four are assigned a topic from the curriculum. Mentors then teach them the mathematics and intuition, then they teach it back to the cohort. Week by week, all associates build the machine learning core together.
The project
Application first. Alongside the theory, every team proposes and builds a data project of their own. They will be mentored from proposal to working demo, and presented to the DAP community at the end of the semester.
No prior machine learning experience needed — only your full commitment.