Jimin Huang is the CEO and COO of The Fin AI, a committee member of The Fin AI working group, and an associated member of The National Centre for Text Mining (NaCTeM). Jimin completed his bachelor’s and master’s degrees at the Computer School of Wuhan University. His research is focused on natural language processing (NLP) and information retrieval.
Title of Talk: Towards AI Readiness in Specific Domains: Benchmarks, Models, Agents, and Applications
Large Language Models are transforming specialized domains such as finance, law, and medicine, but their effective deployment requires dedicated benchmarks, specialized models, and robust evaluation frameworks. In this talk, I will discuss four key areas critical to AI readiness in these fields and The Fin AI’s exploration in advancing them. First, I will introduce FinBen, a comprehensive benchmark designed to assess LLMs across 42 datasets and 24 financial tasks, incorporating agent-based trading and retrieval-augmented generation evaluations. Next, I will discuss Open-FinLLMs, a suite of domain-adapted models optimized for financial NLP, forecasting, and decision-making. I will then present FinCon, a multi-agent system that enhances financial reasoning and collaborative AI-driven decision-making. Finally, I will highlight HARMONIC, a privacy-preserving data synthesis framework that addresses challenges in secure and responsible AI deployment. Throughout the talk, I will showcase The Fin AI’s efforts in these areas and discuss the future of domain-specific LLMs, emphasizing multimodal learning, multilingual adaptation, and AI-driven agents for real-world applications.