Founding engineer and full-stack systems builder working on AI-native collaboration, knowledge stewardship, and practical tools for people doing complex work.
My current work centers on Mixtape, and integrated suite of tools including Crossroads, Catalyst, and Tapestry: related pieces of a production software environment for groups, organizations, and independent practitioners who need to collaborate, produce, and turn dispersed information into knowledge they can actually use.
Crossroads is a collaboration and publishing environment for people and groups. It brings together writing, dispatch, curation, member and group spaces, messaging, public/private surfaces, and AI-assisted work patterns inside one coherent product.
Catalyst is a knowledge-practice and Codex-building system. It helps organizations take the material they already have - documents, spreadsheets, meeting notes, procedures, recipes, research, drafts, and institutional memory - and turn it into structured, reviewable, portable knowledge.
Tapestry is the opt-in relational layer. It is where people and groups can place themselves in a broader field of discovery, public presence, and community connection when that is useful to them.
The shared premise is simple: AI is most useful when people still decide what matters. The work is not just to store more information, but to help people build knowledge they can trust.
The main production repositories are private and available for serious review by request:
mixtape-release-core- Django/DRF backend, PostgreSQL, Celery/RabbitMQ, service auth, operational tooling, Catalyst ingestion, and domain models.mixtape-release-frontend- Next.js/React application surfaces, Crossroads public pages, Catalyst UI, writing workflows, mobile/shared packages, and design-system work.
The broader system includes FastAPI services, semantic-memory work, Qdrant/vector retrieval, Socket.IO/Yjs collaboration, React Native/Expo, VPS deployment, and a governing documentation practice called Puddlejump.
Recent Catalyst work focused on resourcefulness: moving from brute-force document parsing toward human-curated, section-directed ingest, with major token savings and better reviewability.
Before Mixtape, I spent many years building web, data, and public-sector systems across Python, Typescript, SQL, Java, PHP, shell, Solr/search, Tailwind, migrations, CMS architecture, healthcare claims, philanthropy, and nonprofit/small-business software.
I still care a lot about plain old data architecture: what exists, where it lives, who owns it, how it changes, and how people know whether it can be trusted.
Portland, Oregon
marqpdx@gmail.com