MATLAB Agentic Toolkit

The workflow: This open-source GitHub repository introduces the MATLAB Agentic Toolkit, explains how to install and configure MATLAB skills, and helps users build agentic AI workflows.

The user problem: AI coding assistants (like Claude and Codex) are much more useful for engineers and scientists when they understand how to write, test, and run MATLAB code properly. The MATLAB Agentic Toolkit gives AI assistants ready-made MATLAB skills so users can automate engineering tasks with natural-language prompts.

My impact: I restructured and edited the repository documentation, improved the skills catalogue, and simplified the setup instructions so users can get started quickly.

MATLAB in Containers

The workflow: This landing page explains how to deploy MATLAB in Docker containers across desktop, server, and cloud environments, using either prebuilt images or custom setups.

The user problem: Running software on different computers often causes setup problems because each machine has different settings. Containers solve this by packaging MATLAB with only the exact libraries it needs. With containers, users can run the exact same MATLAB setup on a laptop, a server, or the cloud, and plug MATLAB directly into automated testing (CI/CD) pipelines.

My impact: I reorganised and restructured this entire documentation category in collaboration with developers and customer-facing engineers. By grouping content into clear, goal-based sub-categories and incorporating customer feedback, the page now guides users straight to their specific workflow without overwhelming them on a single page.

Multi-GPU Deep Learning in the Cloud

The workflow: This tutorial shows how to train a deep learning network in MATLAB on an Amazon EC2 cloud instance using MathWorks Cloud Center. It walks through training on a single GPU, scaling up to multiple GPUs, adjusting training settings (like mini-batch size and learning rate) for multi-GPU speedups, and saving trained models to Amazon S3.

The user problem: Training deep learning models on a personal laptop can take hours or days. By renting powerful cloud computers with multiple graphics cards (GPUs), users can train large models in a fraction of the time without buying expensive hardware.

My impact: I ran the multi-GPU experiments in the cloud myself and translated concepts from deep learning, GPU hardware, parallel computing, and cloud storage into simple, step-by-step instructions.

Updating & Customising MATLAB Containers (+ Video)

The workflow: This guide shows how to install updates, toolboxes, and add-ons inside a MATLAB container (either automatically with a Dockerfile and MATLAB Package Manager, or interactively through a remote desktop using VNC) and how to save the updated container for future use.

The user problem: Prebuilt containers give users a fast starting point, but teams often need extra MATLAB toolboxes, updates, or add-ons for their specific project, and they want to save those changes so they do not have to reinstall them every time.

My impact: Alongside writing the guide, I created an end-to-end documentation video to walk users visually through this workflow. I wrote the script, recorded my screen, and edited the final video from start to finish.

Creating a Custom Cloud Image (AMI)

The workflow: This workflow explains how to start from a MathWorks base image, install custom libraries, drivers, or toolboxes, and generate a reusable Amazon Machine Image (AMI) for MathWorks Cloud Center.

The user problem: When teams launch cloud computers regularly, installing the same custom toolboxes, drivers, and datasets from scratch wastes time. Creating a custom AMI lets users save a ready-to-use snapshot of their cloud computer and launch new machines from it in minutes.

My impact: I tested the workflow hands-on and worked with developers, editors, and a UX writer across several review rounds to turn a complex cloud-administration process into simple, easy-to-follow steps.

MathWorks Cloud Center UX & Documentation

The workflow: This documentation introduces MathWorks Cloud Center, a web application that lets users launch, manage, and connect to MATLAB and Simulink on Amazon Web Services (AWS) cloud resources.

The user problem: Setting up cloud infrastructure manually requires configuring virtual machines, networking, storage, and security rules. Cloud Center simplifies this so engineers and researchers can start running MATLAB in the cloud in just a few clicks, while clear interface text helps prevent costly cloud-resource mistakes.

My impact: I contributed designs, customer-journey maps, user feedback, and interface text (tooltips, warnings, and cost/deletion confirmations) to the UX of Cloud Center, wrote its documentation, and use usage analytics to keep cloud content focused on what users need.