ClearPath Agent: Multi-Agent Investment Committee

Dec 1, 2025

ClearPath Agent is a multi-agent system designed to replicate the deliberation of a human investment committee. A Supervisor agent coordinates specialized Research, Quant, and Risk agents through ReAct-style reasoning, so that each recommendation is grounded in evidence, quantitative analysis, and an explicit risk assessment rather than a single model’s opinion.

Across a 120-task evaluation, the committee structure improved reasoning accuracy and risk awareness by up to +16% over consensus baselines. The project was built as my Duke M.S. capstone in Fall 2025, where I led a 6-member team across system architecture, the evaluation suite, and paper authorship. It was awarded Outstanding Capstone (1 of 2) for bridging industry and academia, and the coordination protocol was developed further in the FinCom paper.

Multi-Agent Systems LLM ReAct Finance Evaluation
Chao Péter Yang
Chao Péter Yang
Machine Learning Researcher

My research focuses on LLM alignment, agentic systems, and structure-aware generative modeling. I build preference-aligned language models, reliable multi-agent systems, and symbolic music generation models, and aim to bridge theory and practice to create both scientific and real-world impact.