Projects and work

Selected work applying the framework in the real world.

AI Transformation Practicum — hands-on LLM engineering

Ongoing since early 2025: a deliberate full-time investment in hands-on AI skills. Fine-tuned Llama 3.1 (8B) with QLoRA — dataset design through evaluation — exported to GGUF and deployed locally via Ollama; tensor-level model inspection; evaluated fine-tuning tooling (Unsloth, Hugging Face PEFT/TRL, Axolotl) against cost and use-case fit, including when RAG or prompting beats fine-tuning.

AI Trading Agent — built with agentic workflows

Current project: an AI-driven trading application for market analysis, built by directing autonomous agentic coding workflows (Claude Code, Cline, VS Code + LLM) in a product-manager role — from code generation through deployment. Includes a 101-lesson education hub covering options, strategies, and quantitative methods.

AI governance with the NIST AI Risk Management Framework

Applying the NIST AI RMF to govern emerging AI-driven capabilities inside delivery pipelines — governance protocols that keep AI adoption accountable without stalling it. Grounded in federal delivery experience, and reflected in TERRAIN's built-in governance structure.

TERRAIN AI at the Center for Applied AI, UMBC

Invited lunch-and-learn at UMBC Training Centers' Center for Applied AI — introducing the TERRAIN AI Framework to practitioners building AI-enabled organizations: the seven phases, the methods behind each, and how governance and organizational memory keep innovation accountable.