Skip to content
View FromArkZoo's full-sized avatar

Block or report FromArkZoo

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
FromArkZoo/README.md

James Browne

15 years leading enterprise relationships at Bloomberg.

I'm genuinely interested in how frontier AI can democratise knowledge and increase individual agency in the world. The work below is what I've been building in my spare time to explore that question — prototypes, research, and production-shaped tools across frontier AI research, applied AI in finance, regulated-domain data integration, hardware-facing ML, and native iOS reference apps for working professionals.


Move 37

I write Move 37 — a Substack on AI diffusion, frontier model dynamics, and what happens when knowledge work gets compressed by a factor of ten.

Read here →


Research

Genesis Engine — Evolving original abstract strategy games playable by AI agents, using LLM-as-judge evaluation across multiple Claude agent teams. Current direction: cellular automata as the rule substrate for evolvable games. Iteration across 20+ evaluation campaigns, with each generation's failure modes fed back into the constraints of the next.

Applied AI × finance

t212-advisor — A Trading 212 portfolio advisor with Claude-powered analysis, rebalancing suggestions, dividend optimisation, and what-if scenarios. Python + Streamlit + Anthropic SDK.

Data integration in regulated domains

dna-analyzer — A Flask app that cross-references AncestryDNA files against ClinVar, GWAS Catalog, and PharmGKB clinical annotations, wrapped in an editorial-magazine UI (hero photography, display typography, frosted cards). Privacy-by-design: raw files are parsed in-memory and never written to disk.

Hardware × ML

q10-drone-controller — Reverse-engineered WiFi protocol, OpenCV video pipeline, and PyTorch behavioural-cloning autopilot for the AVIALOGIC Q10 mini drone. Packet capture → protocol decode → handshake replay → autonomous flight. Packet captures and replay tooling committed as evidence of the reverse-engineering work.

tv-streamer — An LG webOS homebrew IPTV aggregator with multi-dimensional channel classification (category / language / region), HLS playback tuned for TV hardware, and a CDP → Luna bridge for scripted sideload. Packaged as a .ipk and running on-device.

CycleJames-iOS — Native SwiftUI port of my web cycling-trainer app, built because iOS Safari blocks Web Bluetooth and the PWA can't pair with smart trainers on iPhone. CoreBluetooth FTMS Indoor Bike Data parsing, ERG-mode control-point writes for resistance targeting, Swift Charts ride visualisation, NP/IF/TSS power metrics, SwiftData ride history. Shipped on the App Store (v1.1).

Systems + engineering fundamentals

VoiceChangerPro-iOS — A low-latency iOS voice changer built on AVAudioEngine with FFT processing via the Accelerate framework.

BeanNote — A canvas-based slide / presentation builder in React + TypeScript with a dynamic background renderer.

Reference

JB Glossaries — A suite of three native SwiftUI iOS reference apps sharing one engine: JB Pharma (499 oncology and drug-mechanism terms), JB AI (785 model-architecture, training, and hardware terms), and JB Law (836 US-law terms across 14 categories and four reading lenses). Italic display typography, two-axis indication × category filter, full-text search across term, abbreviation, and definition. Originally a single Python tkinter pharma prototype, now ported and expanded into a multi-industry SwiftUI suite.

Published work

What was that BANG?!? (Troubador, 2026) — An illustrated science book for ages 8–12 exploring how sound travels from physics to conscious perception, built around Socratic dialogue between a child and his father. I also designed and built the marketing site. ISBN 978-1-80634-016-3.


How I work

  • I learn by shipping real things, not by reading about them
  • Evaluation methodology matters as much as the artifact itself
  • Privacy and data handling should be design decisions, not afterthoughts
  • Unknown systems are opportunities to understand deeply — not problems to route around

What I'm thinking about

  • AI diffusion dynamics and the economics of adoption friction
  • Whether the universe is analog or digital (still unresolved, personally)
  • Substrate independence, consciousness, and the ethics of building minds
  • Foundational maths as a hobby and cellular automata

Connect

Pinned Loading

  1. dna-analyzer dna-analyzer Public

    JB DNA — turns an AncestryDNA file into a validated health / pharmacogenomics / ancestry report. ClinVar · GWAS · PharmGKB · gnomAD · dbSNP, plus Google DeepMind AlphaGenome + Ensembl VEP/AlphaMiss…

    Python 1

  2. scratch-mcp scratch-mcp Public

    An MCP server that lets an AI agent build and run real Scratch projects — readable text compiled to a runnable .sb3 (full Scratch 3 block palette, fail-loud) and driven in a live Scratch editor.

    TypeScript

  3. JBGlossary-iOS JBGlossary-iOS Public

    JB Glossary — a native SwiftUI iOS reference app with six industries (Healthcare, AI, Finance, Law, Real Estate, Insurance) sold as IAPs, sharing one engine. 4,732 terms across six shipped books. T…

    Python

  4. aigame aigame Public

    Can evolutionary search discover a board game as deep as Go? Evolving abstract strategy games — PPO self-play, judged by AI agent teams that actually play them.

    Python

  5. t212-advisor t212-advisor Public

    Trading 212 portfolio advisor — Python + Streamlit dashboard with Claude-powered analysis, rebalancing suggestions, and dividend optimisation

    Python

  6. tv-streamer tv-streamer Public

    LG webOS homebrew IPTV app — unified stream aggregator with EPG, M3U parsing, and category/language/region classification

    JavaScript