AI/ML Engineer · Evaluation & Adversarial Robustness · Production ML Systems
Building systems that work under real-world constraints — from edge ML in defense environments to autonomous trading infrastructure and LLM safety evaluation.
Currently an AI/ML Engineer at VES Solutions supporting USSOCOM's Joint Acquisition Task Force.
Automated Trading System — Walk-forward validated momentum bot (Sharpe 5.85, 11/11 OOS windows). Systematically evaluated 31 strategy configurations. Python, vectorbt, Alpaca API, Docker, Ollama.
LLM Agent Red Team — Multi-tiered security evaluation of a RAG agent: naive → hardened → defense-in-depth, with comparative metrics across 5 attack categories. (In progress)
AI Health & Weight Loss System — 4 MCP servers for Claude Desktop: macro tracking, Renpho scale integration, workout data via Hevy API, Telegram bot for daily logging. FastMCP, SQLite, Docker.
Stock Research & Risk Scoring — Composite scoring engine (quality, value, growth, financial health, analyst sentiment) with weekly Telegram reports and HTML risk dashboards.
Trading Portfolio MCP — Model Context Protocol server for managing Schwab positions, portfolio summaries, and position history via Claude Desktop.
- ML systems evaluation and adversarial robustness
- Edge AI deployment in constrained environments
- Production infrastructure (Docker, MCP, local LLM inference)
- Systematic backtesting and strategy validation
📍 Tampa, FL · neetium.com · LinkedIn · Medium · Substack