Full-Stack Engineer
Building production AI agent systems and backend infrastructure.
I'm a pragmatic problem-solver focused on shipping solutions that work. I ship quickly but not recklessly—I follow best practices, document as I go, and dig into optimization when deadlines allow. When I join a team, I immediately map out how to run the codebase, its structure, and where things live, then document it with AI-optimized subject matter trees so future developers don't have to figure it out from scratch.
I started this practice when learning Go in 2024 with Claude, building ModulaCMS—a production-grade headless CMS with concurrent server architecture. To deepen my computer science skills, I developed a data structure learning system that generates practical problems and progressively combines structures based on familiarity. Building this taught me advanced prompt engineering, self-healing workflows, and output validation. I applied these skills at Bonsai, creating a multi-agent orchestration system that generates production code and enables contractor teams to work independently. While I'm full-stack, my passion is backend work—designing APIs, building concurrent systems, and solving architectural challenges.
Outside of code, I'm into rock climbing, producing music, and loving my cats—one of whom has self-appointed herself executive code reviewer and has claimed the spot by my monitor as her office. Before development, I was a recording engineer. Working with bands and artists taught me how to collaborate creatively with different types of people, break down complicated tech into easy to understand language, and distill a creative vision into a technical solution. These skills translate directly to collaborating with designers and product teams today. I work best in environments that value initiative, technical depth, and pragmatic execution.
Claude's WebFetch sends entire raw HTML as input tokens, including every ad and navigation menu. A 100KB documentation page costs 25,000 tokens before Claude reads a word. I built a tool that cleans the HTML locally first, saving 96% of tokens and potentially $43K/month for mid-sized teams.
I watched Claude flash three huge blocks of grep output during a refactor. All that formatting designed for humans, and Claude was paying tokens for colors it can't see. I built a search tool optimized for AI, 4.4x fewer tokens.
Claude kept skipping the npm install step, causing a 30% failure rate on automated problem generation. I tried rewording the instruction, making it more explicit, but nothing worked. Then I realized I needed to ask Claude to confirm it did the step, not just do it, and that changed everything.
FeaturedToken-efficient file search tool with JSON output and MCP server support
FeaturedThis repository is a comprehensive learning path for mastering data structures and algorithms through deliberate practice. It features 33 carefully sequenced data structures, from foundational concepts like arrays and linked lists to advanced topics like suffix trees and bit manipulation, each with practice problems organized by difficulty level.
FeaturedModulaCMS is a fully modular content management system server written in Go that provides content management for both your client site and your admin panel. Learn how in the repo documentation.
Have a project in mind or want to discuss an opportunity? I'd love to hear from you.