The rapid development of cloud computing has attracted a diverse migration of applications. Serverless computing, with its abstract resource management, on-demand billing, and dynamic scaling, has become a popular cloud computing paradigm. X86 signifies traditional computility, while RISC-V symbolizes new potential. Making the most of existing traditional computing resources while exploring the potential of new computility will be a key challenge. The x86 and RISC-V hybrid computility supply will continue for a long time. Task management under diverse instruction set architectures is a critical issue that needs addressing in this evolving landscape. Current research predominantly concentrates on homogeneous instruction set clusters. In this paper, we propose RFaaS, a function job scheduling methodology tailored for RISC-V + X86 heterogeneous instruction set clusters, leveraging the OpenFaaS serverless computing platform. We delve into the affinity traits of function jobs in RISC-V + X86 amalgamated instruction set clusters and devise an affinity classifier alongside an architecture-aware scheduling algorithm. Our methodology dissects scheduling decisions into resource fulfillment and affinity alignment, underpinned by a meticulously crafted update algorithm to uphold and refine job affinities. Experimental results show that RFaaS can provide at least \(\text{3x}\) performance improvement and \(\text{2.4x}\) throughput increase compared to existing solutions.