Neuromorphic computing is emerging as an important frontier of Artificial Intelligence (AI), helping overcome the energy costs and latency limitations of traditional Deep Neural Networks (DNNs) by drawing inspiration from biological neural circuits. In this context, Spiking Neural Networks (SNNs) employ biologically inspired neuron models that span different levels of abstraction, ranging from highly detailed but computationally demanding formulations such as the Hodgkin–Huxley model to simplified and more efficient approximations such as the Leaky Integrate-and-Fire (LIF) model. SNNs encode information through sparse spike events and temporal dynamics, enabling efficient processing of sparse data and naturally supporting temporal coding. This thesis addresses the challenge of bridging the gap between computational neuroscience and specialized neuromorphic hardware.