🎥 Recorded live at the MLOps World | GenAI Summit 2025 — Austin, TX (October 8, 2025) Session Title: SLMs + Fine-Tuning: Building the Infrastructure for Multi-Agent Systems Speaker: Mariam Jabara, Senior Field Engineer, Arcee AI Abstract: As enterprises push AI systems into production, the limitations of massive, general-purpose LLMs are becoming clear — high costs, complex infrastructure, and data security risks when sensitive information leaves controlled environments. In this lightning talk, Mariam Jabara from Arcee AI shares how Small Language Models (SLMs) provide a more efficient, secure, and domain-optimized path forward. She explores the process of fine-tuning and deploying SLMs, including insights from releasing a new small foundation model designed for multi-agent workflows. This session highlights how SLMs can outperform larger models in real-world applications, enabling faster, more reliable, and cost-effective AI systems. What you’ll learn: • Why SLMs outperform LLMs in enterprise-scale multi-agent systems • How fine-tuning SLMs enables domain-specific intelligence and control • The security and latency advantages of running models locally • Practical takeaways for balancing performance, cost, and reliability in MLOps.