I'm Alex Fourie — founder turned CTO. I started WeFix in 2007 and grew it to 35 stores, ~450 staff and a million customers before exiting. Since then I've co-founded a blockchain quant fund, led technology at two companies, and shipped a live online-counselling business, a geospatial marketplace and a fleet of trading systems. I still write most of the code myself.
Alex Fourie · CTO
Fifteen-plus years of building, scaling and exiting businesses, and nearly five as a CTO. I pair board-level commercial judgement — funding, finance, legal, shareholders — with the technical depth to set the architecture, review the code and ship it myself.
Today I lead technology at UbuntAI, an AI consultancy building agents for businesses, and operate TherapyNow, an online counselling practice with clients in 15+ countries. On the side: production trading systems and a drawer full of experiments. I like problems where the feedback is real and fast — a booking that converts, a fill that lands, a store that opens on time.
Seven companies founded or led. One national retail chain. One fund. Two CTO seats.
AI consultancy: agents that answer customers, process documents and run operations for businesses. Also built and operate TherapyNow, an online counselling practice with clients in 15+ countries.
Led technology for a blockchain business — architecture, product delivery, hands-on Solidity and Python.
Blockchain-based quantitative investment fund. Designed, built and ran the fund's trading strategies.
Built a national device-repair chain from zero to 35 mall stores, ~450 staff and 1m+ customers. Owned finance, funding, leases and shareholders. Exited to shareholders in 2018.
Manufactured phone-charging machines and shipped them to 25+ countries; sold the business to WeFix.
Organic electronics-accessory brand stocked by major retailers.
South Africa's first digital music download store; grew the catalogue to 500k+ tracks.
Real things with real users — designed, built, deployed and operated. Mostly serving the South-African market.
The AI consultancy I lead as CTO. We build agents that answer customers on WhatsApp, process documents and run operations for businesses — human hand-off and audit logs as standard. The site runs its own Claude-powered lead assistant.
ubuntai.clickA two-sided local services marketplace connecting providers ("Bustlers") with buyers, powered by real geospatial search — Postgres + PostGIS distance queries, full-text ranking and realtime messaging. The most architecturally mature thing I've built.
An online-counselling business — 16 registered SA therapists covering trauma, anxiety, depression, couples and addiction, with a "book in 60 seconds" flow, automated reminders and corporate plans.
therapynow.co.zaAn end-to-end AI content factory: scan the news → write lyrics → generate music → generate imagery → compose video → publish. A worker-based pnpm monorepo orchestrated with the Anthropic SDK and ffmpeg.
A conversion-focused sales page for a premium crypto-futures signal service delivered to Telegram — the front door for the trading systems below.
An SEO lead-generation microsite for satellite internet in South Africa — home, farm, business and off-grid landing pages engineered to capture rural connectivity demand.
Automated systems across crypto, forex and prediction markets — engines, scanners, backtesters and dashboards. Built and operated; described as engineered systems, not guaranteed returns.
My most operational system — a Python forex trading engine with a live loop, SQLite trade ledger, Parquet logs, reconciliation jobs and a watchdog, all dockerized with a restart runbook.
A 5-minute Bybit futures bot on a mean-reversion + momentum-confluence strategy (ADX, MACD, VWAP, RSI divergence) with ATR-based dynamic stops, stepped trailing, a scanner, dashboard and Telegram alerts.
An AI-driven USDT-M futures bot that scrapes Binance Square social feeds for hype, asks Claude for a trade decision, and enforces a hard per-trade loss cap.
A Polymarket bot that arbitrages crypto-price prediction markets against Binance spot across multiple pairs, with named strategy variants and a backtesting harness.
A Bybit liquidation-hunter that fades large forced liquidations, filtered by open-interest share, with server-side ATR trailing stops, paper + live modes and an equity kill-switch.
Tracks all 157 tokenized-stock pegs on Mantle (e.g. TSLAx vs NAV) twice daily to surface de-peg risk and arbitrage gaps — a keyless data core backed by Dune queries.
A Flask dashboard scanning exchange open interest and funding rates to flag crowded positioning and funding extremes across the market.
A scanner + backtester + dashboard codifying three small-cap short strategies (gap-up, bounce, first-red-day), with every rule cited to source — runs fully on synthetic data, no keys required.
链上雷达 — a Flask dashboard plus background scanner surfacing emerging crypto narratives and on-chain signals in one place.
The R&D drawer — SaaS scaffolds, niche utilities and a couple of genuinely strange games. Where the next products get prototyped.
A portfolio of 8 standalone FastAPI + Claude micro-SaaS APIs — code review, doc generation, tutoring, health and more — each with a free tier and PayPal billing.
"Every stay, secured" — SaaS for short-term-accommodation contracts and breakage-deposit holds, where agents send guests e-sign contracts and deposit authorizations.
Issues small-vessel buoyancy / seaworthy certificates per SAMSA marine regulations in minutes — a Next.js tool that renders official PDFs and emails them out.
"The Burning Throne" — a last-man-standing on-chain game on Solana, playable by humans and AI agents alike. An Anchor program with a one-button Next.js frontend.
A darkly-comic trading life-simulator — you versus three AI opponents, with bankruptcy played for spectacle. A browser prototype in Vite + TypeScript.
A Claude-managed memecoin paper fund — the model runs a virtual portfolio against a JSON ledger, with an HTML dashboard to watch it think and trade.
I take an idea from a blank folder to a deployed, operated product — design, frontend, backend, infra and the boring bits like reconciliation and runbooks.
Claude and LLM agents are woven directly into what I build — from trade decisions and content pipelines to the lead assistant on UbuntAI's site.
Bias to launch. Real users and live markets teach faster than planning docs — so I get things into the world and iterate from what actually happens.