The success of Artificial Neural Networks (ANNs) trained with error-backpropagation algorithms relies on the combination of offline training, vast amounts of data, and energy-intensive simulations on dedicated hardware accelerators. In contrast, biological neurons can learn quickly in a self-supervised manner by utilizing (a) recurrent connections within their local network and from other areas, (b) segregated dendrites receiving inputs from feedforward and feedback pathways, and (c) learning rules that combine locally available information from pre- and -post-synaptic sources. Inspired by local learning mechanisms observed in the brain, a plethora of spike-based learning circuits have been developed to train SNNs [1,2,3]. These algorithms are particularly useful for application on neuromorphic devices where signal processing and learning are constrained similarly to biological networks. Indeed, neuromorphic devices show low-latency and low-power computational capabilities by sharing information with sparse, event-based encoding, and combining information storage and processing on low-precision heterogeneous synapses [4,5]. However, these bio-inspired mechanisms are limited in their expressivity and computational power. Importantly, they support efficient credit assignment only for shallow networks, resulting in limited applicability for real-world problems. A key challenge remains how to effectively train SNNs for neuromorphic hardware implementations. To address this challenge, we propose an innovative learning framework for SNNs that combines spike-based local learning [4,5] and control feedback signals [6]. The algorithm operates locally and does not require backpropagating gradients, allowing online learning in SNNs and Neuromorphic devices. First, we apply the algorithm to train SNNs and test it on various dynamic datasets. Second, we validate the hardware implementation of our algorithm with behavioral simulations of neuromorphic circuits. Our results indicate that our algorithm allows for efficient and online training of SNNs despite device mismatches, and allows In conclusion, this study introduces a novel learning framework for SNNs that strengthens the links between biologically inspired learning and neuromorphic computing. The empirical validation of this method across various datasets and its effective behavioral simulations on neuromorphic hardware demonstrate its potential to revolutionize the scalability of neuromorphic devices.