THE STORY

Experience taught me what matters. Building taught me what’s possible.

My 15 years of helping tech companies navigate GovCon inspired me to dive hands-on into AI engineering, driven by a single mission: using technology to unlock human potential.

Abraham Xiong wearing a royal-blue shirt and fedora33.7490° N · 84.3880° W
“Living my gifts, helping others walk their journeys.”The why behind everything I build

THE WORK BEFORE THE TECHNOLOGY

The lesson was never just how to win a contract.

I’ve spent 15 years figuring out why some small businesses win government contracts and why most do not. Along the way, I helped businesses secure more than $35 billion in contracts and generate over $2 billion in revenue.

The deeper lesson was about access: how knowledge, relationships, systems, and confidence can change what an organization believes is possible. My work has always centered on creating more of that access.

THE MISSION

Govafy put my domain experience into software.

Govafy began with a problem I knew from government contracting: useful knowledge is scattered across award data, documents, people, and decisions. I architected the platform to connect those sources and make them easier to work with.

The project brings together TypeScript, data ingestion for more than 30 million award records, OCR for documents, and RAG with source traceability. Its case study explains the architecture, my contribution, and what building it taught me.

Project Details

THE TECHNOLOGY BORN INSIDE IT

To build Govafy, I had to reinvent how I built software.

I had not coded since high school, but I needed to ship enterprise-grade software with confidence. A standard single-agent process was not enough. I needed a system that could plan, specify, build, test, review, and ship.

I created a meta-harness that runs tournaments of competing agents from different frontier labs, then fuses their strongest results. That process became Fusion Coding. A shared specification lets me compare implementations against the same requirements, select or integrate appropriate approaches, and verify the result.

I use Linear to track the work and Claude Code, Codex, and Cursor throughout development. I define the requirements and architecture, review implementations, test and debug the results, and manage integration and deployment. The tools help me build; the decisions and responsibility remain mine.

Project Details

TWO PROJECTS · ONE MISSION

Govafy shows the platform I architected. Fusion Coding shows how I approach building and verifying software with AI.

These projects have moved me from describing a problem to building and evaluating a possible solution. I can now have more informed conversations about architecture, implementation, and deployment, and contribute directly to the software behind those conversations.

See what I’m building