Built and deployed an API for BlockScience’s Knowledge Management System used to organize and manage internal documents, search institutional knowledge, and support an intelligent ChatGPT integration.

Developed a specification and reference implementation of “Reference Identifiers (RIDs)”, a system for representing and interacting with arbitrary knowledge objects in the context of a graph representation.

Documented and presented findings with the BlockScience internal team and larger community.

The Telescope bot is a tool which assists ethnographic researchers in creating datasets and automating the consent approval process within online communities (currently on Slack, it previously supported Discord). It is a participatory process which acts within the field site where a study is situated, allowing community members to see when their messages are requested, manage their consent, and view a live feed of released comments. This approach respects organizational boundaries, making it explicit when data leaves its native environment. Furthermore, the bidirectional flow of knowledge increases trust between the host and research organizations. A KOI-enabled Obsidian plugin allows researchers to integrate the live feed of approved data, including retractions, anonymizations, and on-site notes, directly into their personal knowledge management systems. The same network allows researchers to share live versions of their notes into each other’s Obsidian vaults. An experimental integration shares researcher notes back to the host Slack community, supporting a complete cycle of knowledge processing: collection, consent, synthesis, and publication back to the space where responses can once again be collected.

Developed and maintained a Discord bot written in Python used by RMIT’s Crypto-Governance Observatory to gather 183 messages from 5 communities via a 3-step consent process.

Data collected using this tool and the results of the ethnography published in the Qualitative Inquiry.

Developed a networking app for users at the DWeb Camp conference to easily exchange personal information by scanning a QR code and add to a global graph visualization of connections made.

Deployed a full stack web app using Nginx, Uvicorn, FastAPI, Neo4j and Bootstrap, and configured a portable server to run on the conference’s local mesh network

Presented my findings from the project after organically acquiring over 150 users.

Worked with authors of the EIP-4824 DAO interface standard to implement the specification for 8 blockchains as a robust REST API service using an nginx, Gunicorn, and Flask back end and a custom Bootstrap + JS front end

Used AWS DynamoDB and IPFS to provide scalable and decentralized storage of DAO metadata.

BoxQuest is an experimental social platform and interface for the physical world focused on real interaction, social connection, and a sense of place.

Users interact with physical objects (currently cards) that are tagged with a unique QR code. Scanning the code lets users access a thread associated with the object and add their own entry to it. Each entry is composed of an image, geotagged location, and message. The thread shows the path that the object took through time and space to get to its current holder. By only allowing the current holder to post in the thread, the object passes from person to person, each adding their own entry marking its spot in space and time.

On top of this system, each thread also comes with a goal or topic, called a “quest” attached to it. The first person to use the tag decides its quest and writes instructions on how to pursue it. Possible quests could range from “take a picture of your pet” or “pick up litter” to “send this card around the world and bring it back to me” or “do a good deed and pay it forward”.

This concept was inspired by Karl Schroeder’s “The Suicide of Our Troubles” in which a mixed reality game begins to catalogue and incorporate real world objects and places. The system eventually gives rise to AI-powered characters who seek “suicide” by resolving the environmental issues they represent.

I had originally imagined an approach similar to this story, in which gamification and tokenomics could incentivize a decentralized network of users to perform physical actions on behalf of other users or algorithms. However as I started to build out the prototype I realized how reductive this model was, which could easily lead to another exploitative gig work platform. If instead BoxQuest could socially organize people by building interest and connections to the physical world, it could be an effective force for good!

BoxQuest is a genuinely new spatial architecture for social media, and for once “space” actually means space. I found that social interaction itself can be powerful and build real networks. The physical exchange of tagged items, sense of communal investment in a quest, and positive engagement it encourages aren’t present on any other platform. Furthermore, this approach also naturally facilitates organic growth of the platform. Each time a tag is passed on, a new user is introduced to BoxQuest and instructed how to use it.

I’ve been dreaming up this concept for a long time and am really excited to share this with everyone at DWeb Camp. I can’t wait to see what quests people come up with and where the tags will travel once camp ends!

Developed an implementation of Modular Politics, a system for modeling a wide range of governance structures.

Created a flexible user organization and voting system, 20 nestable modules for governing in-game actions, and support for developers to write their own modules in Lua for the open-source sandbox game, Minetest.

Prototyped a modular software model in Python for creating and maintaining complex multi-step contract systems via REST API requests and used it to build two functional demos spanning three platforms.

Coauthored a paper on “net-native agreements,” published in the MIT Computational Law Report.