Miles Thompson

Technologist, Sociologist, Optimist

Developer, artist, programmer, social founder, technical architect and founding CTO of the leading independent financial research provider in NYC.

25+ years stellar experience as founder, developer, technical architect, and technical lead.

Currently (2023)

  • Papa Reo: Machine learning and data science for an indigenous led organization building speech, language understanding and synthesis for under resourced languages across the Pacific.
  • Mutual Credit Services: Co-design of 'collaborative finance' solutions with communities and activists delivering social impact.
  • Together Project: An open source project building a high fidelity mobile experience for community engagement.

Formerly

  • Founding CTO at leading NY financial research company CreditSights, acquired 2020
  • Founding CTO at AI / Edutech startup Chatterize, acquired 2022

BSc Hons I, Math

1995, University of Otago, NZ
First Class Honours (Summa cum laude)
Mathematics

BA, Sociology

1994, University of Canterbury, NZ
Bachelor of Arts

Member of Trusted Seed

2021

Co-inventor Medical Patent

2001 patent: US20030061072

Current Projects

Papa Reo

AI Lead Dev

  • Deep learning / machine learning
  • Dev ops, Cloud infra
  • Tensorflow, Pytorch, Python, Kubernetes

Working with iwi group TeHiku.NZ in 2019 I led the machine learning effort to train models and build an API for the world's most accurate automated speech recognition (ASR) system for Te Reo Māori. When presented at language conferences in Oxford and Cambridge senior developers from Mozilla described themselves as "shocked and amazed" at the high accuracy of our model. I am now with Te Hiku full time.

Mutual Credit Services

Technical Advisor

  • Tech architecture, collaborative design
  • Svelte, VueJS, Supabase, Hyperledger, whatever it takes
  • Collaborative economics and user experience design

Mutual Credit Services envision a global Credit Commons underpinning a collaborative economy that has become 'business as usual'. We have a large goal but we're building this one community at a time.

Together Project

Co-founder

  • Open Source / Community Engagement
  • Mobile UX, Flutter JS, Figma, DGraph

Together is a project that asks the question - what if we could actually see the network of connections showing who has been acknowledged by who in a community?

We're building a high fidelity, open source, mobile experience and aim to become the foundation platform whenever people want to build communities based on reciprocity and open exchange.

Recent Projects

Cashless Society

Co-founder / Source

  • Distributed crypto meets mutual credit
  • Business Design, Pitching, UX
  • Eth / Solidity, Scuttlebutt, React

Cashless is about making money human again - merging crypto primitives with a new way to think about mutual credit. We provide an alternative way to keep track of credit through communities and tools for greater reliability and transparency - allowing communities to scale trust and act as their own bank.

Resilience App - Mutual Aid World

  • Open source, community engagement
  • React, Mobile, Heroku, Firebase

Thousands of mutual aid groups sprang into existence in March and April 2020. "Mutual Aid World" built software to support such groups. After co-founding with three others I helped organize hundreds of volunteer developers, designers and contributors of all stripes. As well as active tech support we built cryptographically strong contact tracing, a phone support line, and the "resilience app" - which was a react mobile app for organizing delivery of food boxes. In partnership with local groups we helped deliver thousands of food boxes.

Research, development and whitepaper on community resilience

  • Bayesian modeling
  • Decentralized messaging
  • Community resilience

Research and development into resilient coordination protocols in distributed, unreliable environments. Simulation of information flows in distributed messaging environments. Bayesian modeling of multi-agent systems in the face of variance in the reliability of actors. Most of this work remains proprietary but the 'realtime fact benchmark' is one of it's public outputs.

Deep PBR

  • Machine Learning
  • VR / AR
  • Python, Tensorflow, AWS

Working with the VR pioneers at RealityVirtual I designed, trained and deployed a new machine learning pipeline and API for creating virtual reality assets (displacement maps, normals and the like) from regular SLR photographs.

Paekākāriki Online

  • Community Engagement
  • Web and UX design
  • PHP, Custom Wordpress Theme and Plugins

Working with local community groups I helped coordinate and was the developer for a website to represent the community where I live - Paekākāriki, NZ. This features a services directory, calendar, radio station, forums and information for tourists as well as photo essays, poetry and other content.

A love letter to the town in which I live, produced in collaboration with my community Visit paekakariki.nz

Paeakakariki NZ

Art

Performance (by GPT)

  • GPT4 Experiment
  • Poetry
  • Theater!

A wee experiment from March 2023, using GPT 4 to create poetry. Yes, I let GPT choose the content, write the poem and do a lot of the presentation. I did provide a long prompt to bound the piece, however.

View here - Watch till the end!

The great outdoors: algorithmic theater

  • Text processing
  • Machine learning
  • Theater!

I created the text for Annie Dorsen's 2017 theater piece The Great Outdoors (in collaboration with Marcel Schwittlick).

This work involved building a system to harvest text from recent comments on the internet, measure the textual entropy of that text, and select it out in various ways to fit the aesthetics of the piece. Developing the algorithm collaboratively with Annie Dorsen was a truly fascinating and wonderful experience. Working at this level in theater was another awesome treat for me. I wrote the project up as a talk for a machine learning meetup, the slides of which are below.

View slides from the talk

Glitchaus: Data knits

  • Data processing
  • Glitch Art
  • Fashion

Over some years I have collaborated with Jeff Donaldson, one of the early protagonists of the Glitch-Art movement and creator of the Glitchaus fashion brand on his data-knits project. Amongst other things, I created a tool that he can use to take any binary and prepare it for physical projection into woven or knit textile form. Jeff used the program to prepare work for his Bauhaus collection as well as other pieces.

Check out @glitchaus on twitter.

Resume