This project explored how artificial intelligence can revolutionise the design and optimisation of advanced two-phase microfluidic cooling systems used in high-performance technologies. By combining machine learning (ML), acoustofluidics, and nanostructured surfaces, the research developed a smart cooling platform capable of self-optimising its performance in real time. The project introduced an innovative system that integrates acoustically driven bubble activation with nanoarray-coated micropin structures. Unlike conventional cooling systems that rely purely on passive heat transfer, this active system uses ultrasonic waves to control bubble generation and flow behaviour at the microscale, stabilising boiling and improving heat distribution. The nanoarray coatings enhance capillary action and wettability, allowing the liquid to refill rapidly and maintain stable thermal operation. Machine learning models—specifically Long Short-Term Memory (LSTM) and Deep Neural Networks (DNN)—were used to predict and optimise thermal performance based on experimental data. The models achieved over 99 per cent accuracy in forecasting the heat transfer coefficient, identifying the key factors influencing cooling behaviour: initial temperature, chipset material, nanoparticle type, and flow rate. Statistical tools such as SHAP analysis and Partial Dependence Plots further explained how these parameters interact to shape system performance. The findings demonstrate the potential of AI-enabled simulations in thermal system design optimisation. By coupling data-driven models with physical experimentation, the project provides a framework for intelligent, adaptive cooling systems. These techniques can be applied beyond electronics to AI hardware, data centre management, and micro-reactor design, where precise and efficient heat regulation is critical. The approach showcases how integrating AI and physics-based modelling can accelerate innovation in thermal management, leading to smarter and more energy-efficient technologies.