Artificial Intelligence & Data Science
Building human-centered AI systems through iterative refinement and strategic implementation across enterprise environments.
Building human-centered AI systems through iterative refinement and strategic implementation across enterprise environments.
Philosophy & Approach
Manifesto
AI should augment human intelligence, not replace it. The most powerful AI systems are those that seamlessly integrate into human workflows while respecting values, privacy, and agency.
Core Principles
- Context loading and refinement drives better outcomes
- Simplicity and clarity triumph over complexity
- Iterative improvement beats perfect first attempts
- Human judgment remains essential in AI-assisted decisions
Projects & Experience
Worker Bee AI Coordination
Master of Science in Information Systems capstone project exploring multi-agent coordination patterns inspired by colony intelligence.
Key Outcomes:
- Designed a HIVE-style coordination framework for autonomous AI worker agents
- Researched task delegation, conflict resolution, and emergent group decision-making
- Authored UOnline preprint synthesizing capstone findings for broader publication
Personal Website Rebuild – ColePreece.com
Ground-up rebuild of ColePreece.com featuring comprehensive backend CRM functionality, accessibility-first design, and global data privacy compliance across 7 jurisdictions.
Key Outcomes:
- Shipped a full backend CRM (privacy center, consent versioning, bulk email, analytics) on Next.js 16 + Vercel
- WCAG-conscious component library with accessibility-first interactive elements
- Privacy compliance covering GDPR, UK-GDPR, CCPA/CPRA, Quebec Law 25, plus 4 additional jurisdictions
- AI-paired development workflow using Cursor + Claude for accelerated full-stack delivery
AI SOC Triage Agent
University of Utah Graduate School of Business independent study applying agentic AI to security operations center triage and incident routing.
Key Outcomes:
- Designed an LLM-driven triage agent that ranks and routes SOC alerts by risk and ownership
- Integrated context-loading patterns to preserve analyst decision-making authority
- Evaluated against real-world incident workflows for accuracy and operator trust
Trade with Tricia Founder
Part of E-Scholars University of Portland accelerator program. Envisioned a chatbot that could help students trade textbooks with other students on campus.
Key Outcomes:
- Developed textbook trading chatbot concept
- Participated in university accelerator program
- Gained entrepreneurial experience
- Learned product development fundamentals
Methods & Tools
Methods & Frameworks
Tools & Technologies
Current Challenges
The questions and problems I'm actively exploring and working to solve
Context Management
How do we effectively load and maintain context across complex AI interactions to achieve optimal results?
AI Reliability
What strategies ensure AI outputs remain consistent and trustworthy in production environments?
Human-AI Balance
Where should we draw the line between AI automation and human oversight in critical decisions?