Before I worked in software, I served as an all-source intelligence analyst in the U.S. Army. The job was built around incomplete information. I had to separate signal from noise, understand how complicated pieces affected each other, and explain the practical meaning to a stakeholder who needed to make a decision.

After leaving the military, I spent a few years moving between college programs and trying to find the right direction. The constant during that time was building. I worked on personal projects in data analytics, full-stack software, retrieval systems, model distillation, memory and context management, MCP servers, and APIs.

Those projects became Zebra Digital. Today I work across two overlapping disciplines: full-stack software engineering and AI systems engineering. My work ranges from SaaS applications and internal tools to data pipelines, retrieval systems, model adaptation, agent infrastructure, and evaluation. I take a particular interest in turning a demo or early prototype into a working MVP or beta, and I also step into stalled software that needs to be repaired and moved forward.

I currently work with Consciens as a contractor and core member of its internal team. My work there includes the database, embedding and retrieval systems, caching, context management, API and MCP routes, and connecting those systems to the product frontend.

I use AI coding tools extensively and do not pretend otherwise. They help me move quickly. My responsibility is to make the architecture sound, connect the pieces, test the behavior, document the system, find the failures, and deliver software that can keep moving after the first demo.

How I Approach the Work

Make the complicated understandable

I break systems into clear parts, explain the tradeoffs, and make sure the person making the decision has what they need.

Be honest about the process

AI tools are part of how I develop software. I am transparent about that workflow and accountable for the result.

Move quickly, then harden

A fast MVP creates momentum. Tests, security work, documentation, CI, and deployment support are what make it dependable.

Leave useful documentation

The code should not become a mystery after handoff. I document the architecture, interfaces, setup, and operating decisions that matter.

Tools and Technologies

PythonTypeScriptReactNext.jsAstroFastAPIPostgresSupabaseAnthropic SDKOpenAI SDKChromaDBVercelTailwind CSSMLXQwenDeepSeekscikit-learnpandasLocal inference
Tell me what you're building →