This project developed a practical, AI-enabled framework to quantify and reduce airborne infection risk in offices using simulations and data-driven design. The team built a new probability-of-infection metric by coupling high-fidelity computational fluid dynamics (Eulerian–Lagrangian particle tracking) with models for exhaled CO₂ and “age of air,” which measures how effectively fresh air reaches occupants. The methodology demonstrates how particle emissions, exposure time and clinical viral-load data are combined to generate spatial–temporal risk maps that evolve with time. The study then trained machine-learning regressors to predict infection risk directly from two easily measured inputs: indoor CO₂ concentration and the supplied ventilation rate. In a validated office case, the analysis highlights ventilation effectiveness—not just how much outside air is supplied—as the key lever. Recirculation zones create “hot spots” of old air and trapped aerosols; contour plots on page 10 illustrate how quanta (infectious particles) accumulate where flow stalls, which is not always captured by CO₂ alone. Volume-averaged curves show that, under the tested layouts, infection probability can rise rapidly despite meeting nominal ventilation rates, underscoring the importance of diffuser placement and flow distribution. For the data-driven layer, an optimised random-forests model predicted the CFD-derived infection risk with high fidelity (coefficient of determination ≈ 0.99; low RMSE), enabling a closed-loop concept: smart building systems can infer risk from CO₂ sensors and adapt ventilation in real time without running expensive simulations. This directly supports AI-enabled simulations and data-driven design workflows, where surrogate models and sensor data inform rapid iteration of diffuser layouts, occupancy policies and control set-points. Beyond offices, the approach can feed digital twins for schools, hospitals and public venues, accelerating safer, energy-aware ventilation design while reducing modelling overheads.