What projects can students build in the SAAI Research Program?

Research projects are selected to combine a meaningful question with measurable technical evidence. A student may compare different vectorisers and classifiers, build a sentiment-analysis pipeline, investigate how a model behaves across datasets, create a concept detector, explore local-language or Hinglish text, study classification errors, or develop another mentor-approved AI experiment. The project must include more than a final prediction. Students are expected to define the problem, identify relevant variables, prepare or audit data, choose methods, run experiments, record results and explain limitations. Stronger pathways may also include an API, dashboard or deployable demonstration so that the research can be inspected and used by others. Project scope is adapted to the learner’s level, but every project should produce evidence that can be reviewed rather than relying only on a certificate or presentation.