Experimental Design: Learn to design and run studies end to end — defining hypotheses and outcome measures, assigning intervention and control groups, and reasoning about confounds, attrition, compliance and statistical power in real-world settings where randomisation is imperfect.
Validation of AI-based Measurement: Gain hands-on experience validating LLM-based graders and evaluators against human-rated ground truth, including rubric design, rater calibration, agreement analysis, and diagnosing where automated judgment diverges from expert judgment.
Causal Inference and Statistical Analysis: Strengthen the ability to estimate treatment effects from experimental and observational data, quantify uncertainty, handle non-compliance and dosage effects, and distinguish statistical significance from practical significance.
Research Operations: Gain first-hand experience running studies with real institutions — participant recruitment, consent processes, instrument design, and the practical work of getting a study to happen alongside academic collaborators and school partners.
