About Me

I’m David Duncan, a software engineer working on AI-driven trading systems and Linux infrastructure.

Current Projects

Colosseum - Multi-Agent Investment Framework

I’m building Colosseum, a multi-agent system that simulates an investment committee using AI agents. The framework orchestrates specialized agents (technical analysis, fundamental analysis, sentiment analysis) that collaborate on trading decisions.

Tech Stack:

• LangChain/LangGraph for agent orchestration
• Model Context Protocol (MCP) for secure broker integration
• PostgreSQL for state management
• Podman Quadlets for containerized deployment
• Python 3.9+ with async/await patterns

Current Focus:

• Broker integrations (Interactive Brokers, E*TRADE, DAS Trader)
• Risk management layers
• Backtesting infrastructure
• Agent specialization and coordination

Linux & Infrastructure

I work with Fedora Linux and containerized deployments:

• Podman/systemd integration
• Package maintenance and RPM building
• Container orchestration with Quadlets
• Production deployment automation

What I Write About

This blog documents real technical work:

• AI agent architecture and design patterns
• Secure API integration strategies
• Container orchestration and Linux infrastructure
• Performance optimization techniques
• Actual code from real projects

No hypothetical examples - everything here is based on production systems and real implementations.

Get in Touch

• GitHub: github.com/davdunc
• Email: davdunc@gmail.com
• Colosseum Project: github.com/davdunc/colosseum

About This Blog

Jekyll-powered, GitHub Pages-hosted. All posts reflect real technical work and lessons learned from building production systems. If you have questions about any of the topics or want to discuss implementation details, feel free to reach out!