A mix of open and closed source software I’ve worked on throughout the years. Many projects have links you can follow in order to interact with the software directly

Aleo SDK

Rust, JS/WASM & Python SDKs for Aleo Protocol

aleo

Employer: Aleo

Year: 2023

Description

Creation of Rust, js/wasm & python SDKs to provide accessibility to new protocol features for developers.

Links To Work

Private Cryptocurrency Gifting

Implementing New Transaction Types

mobilecoin

Employer: Mobilecoin Inc.

Years: 2021-2022

Description

Expanding the number of useful transaction types within the Mobilecoin blockchain

It was desired to create the ability to send Mobilecoin to mobile phone users who did not yet have Mobilecoin accounts. A solution was created allowing Mobilecoin account holders to send a transaction to a reserved subaddress and then send the data needed to create an account and redeem that transaction over gRPC to a counterparty without a mobilecoin account

Links To Work


Forest Framework

Private Applications in Signal Messenger

signal_app

Years: 2021-2022

Description

Framework enabling private, commerce enabled applications to be built within Signal

The framework is built from:

  1. Auxin - A server-side Signal client implemented in Rust which allows applications to communicate with humans through Signal. Built using Signal’s libsignal which exposes the cryptographic primitives behind the Signal Protocol needed to build 3rd party signal clients

  2. Forest - An async Python based application framework that allows people to build automated chat applications with AI driven dialog flow and fiat + cryptocurrency payments

Mobilecoin, a cryptocurrency integrated into Signal Messenger acquired the Forest Framework 2021 in order to provide developers with a platform to build Signal apps that supported Mobilecoin transactions

Links to Work

Live Applications

Github Links


Scripd

Automated Prescription Reminders Using NLP and Computer Vision

medical_app

Client: Scripd Inc.

Years: 2020-2021

Description

Scripd is a Flutter based app which allows users to scan their prescriptions and automatically populate prescription reminders. Text is recognized using CNN based Optical Character Recognition (OCR) and prescription data is populated using transformer based Named Entity Recognition (NER) applied to the text recognized by the OCR

Both the OCR and NER Models are run directly within the app to maximize speed and avoid the need for internet connectivity. Updates to the models are bootstrapped in the background when users are connected to WIFI


BlockUSign

Document Signing via Ethereum Blockchain

dash

Years: 2018-2019

Description

Proof of concept for credential verification on Ethereum

BlockUSign takes a hash of an uploaded document, signs it with an Ethereum account’s private key and stores it on the Ethereum blockchain. Any other user can validate that document later using the document and public key of an Ethereum account. If a user is signed in with their Ethereum account, the app will scan the Ethereum chain for the associated documents and display them to the user

This served as a proof of concept for a larger product designed to allow people to prove ownership of sensitive official credentials (such as financial statements or medical records) to counterparties in a private way

Links to Work


AI Support Assistant

Support Request Classification + Useful Language Integrations

loggly

Client: Loggly

Years: 2014-2015

Description

Custom AI layer built into a Salesforce CRM for recommending actions to increase sales conversion and improve responses to support requests. The framework performs the following tasks:

  1. Classify the potential value of prospective clients and route them to support reps with specific sales recommendations
  2. Add customized responses to automated support email and chatbot replies
  3. Classify support requests from existing clients with severity levels and recommendations

Links to Work


PlaySignals

Analytics Dashboard for Social Network Games

dash

Client: Game Titles Developed by Bitfold

Years: 2011-2013

Description

Analytics service for multiplayer mobile games

Measurements of player in-game actions and game state were collected by the UI and warehoused into a Postgres database. Metrics and simple predictive models were built into Postgres stored procedures and run against the data regularly. An EC2 based web-service was used to generate visualizations of the data onto a browser based dashboard for analysts on demand

Several game publishers chose to integrate the dashboard into their games

Links to Work

The service isn’t operational anymore, but the following articles explain the thinking behind the product