Skip to content

Projects

Production systems, and two games built from scratch.

ML and data work inside a scaling food-tech business, and two games that run in your browser. Confidential specifics are kept out.

Games

X-STRANDEDA derelict ship, a hive that hunts by sound, and one flashlight. My own WebGL2 engine, playable in the browser.Enter the dark

Playable in your browser

Child of Anon

An original auto-scroll climber and runner in TypeScript on Phaser 3: jump, double-jump, dash, combat with spells and mana, real pixel parallax, and online VS through a Cloudflare Durable Object. A game I first prototyped in Python, rebuilt with today's tooling.

  • Phaser 3
  • TypeScript
  • Vite
  • Durable Objects
  • Cloudflare Pages

Machine learning & data

Deployed

Demand Forecasting & Food-Waste Reduction

Forecasting models and end-to-end data pipelines that sharpen demand prediction across a scaling catering operation. Better predictions mean better resource planning — directly reducing food waste and improving margins.

  • Forecasting
  • Python
  • Operational Analytics
  • Optimization
Deployed

ERP Integration & Internal Platforms

Contributed to ERP development and maintenance, plus several internal web apps and operational dashboards. Connects operational data to the people who act on it, with workflow automation that removes repetitive manual work.

  • ERP Systems
  • REST APIs
  • Internal Tooling
  • Automation
Deployed

Operational Analytics Infrastructure

Scalable data collection and transformation workflows feeding sales and operational analysis. The analytics backbone that turns day-to-day business activity into decisions about efficiency and profitability.

  • Data Engineering
  • Analytics
  • Python
  • Profitability Analysis
Deployed

End-to-End ML Pipelines deployed in AWS & Docker

Production-grade ML pipelines — from feature engineering through deployment — running on AWS with Docker for environment standardization and repeatable, reliable shipping to production.

  • ML Pipelines
  • AWS
  • Docker
  • Production ML

These represent ongoing, real-world work inside a scaling food-tech / catering business. Want to talk specifics or see how a similar system could help your operation?Get in touch →

How models learn

Descending a loss landscape.

Training is optimization: a model rolls downhill across a surface of error, step by step, toward the valley where its predictions are most accurate. Every forecasting and ML pipeline above is built on this idea.

Engineering & systems

End-to-end builds that show ownership from requirements through deployment and servicing.

Project

Industrial PC Deployment & ERP-Managed Service Pipeline

End-to-end industrial deployment: a custom application on a locked-down industrial PC, serviced through an SFTP-to-S3 workflow and maintained via a custom ERP module — a complete pipeline from business system to deployed operational hardware. Owned from requirements through deployment and servicing.

  • Industrial PC
  • Custom App
  • ERP Module
  • SFTP
  • S3
  • Systems Integration
Project

Dynamic Product Portfolio Web App

Customer-facing React application serving a tailored product-sheet portfolio that updates as backend data changes and adapts to customer preferences and prior interactions — moving product communication from static documents toward a living digital interface.

  • React
  • Backend Integration
  • Product Data
  • Dynamic Web App
  • UX