Lee-Michael D'Souza

Global Kubernetes

Created, upgraded and maintained 10+ global clusters hosted in the cloud (Azure + AWS), each consisting over 1000 Pods with their own Network, Storage and other resources. This included scaling resources horizontally + vertically, handling resource, upgrading k8s and Namespace rollouts, investigating crashes, monitoring, etc.

Created and ran jobs through k8s and also through Python scripts across all Pods. Also built in and managed logging functionality to Grafana through promtail/prometheus/loki.
Led, created and deployed development of new Apps (Python and JavaScript) through Namespaces to expand on existing functionality.

Cloud: Terraform, Azure & AWS

Developed, containerised and deployed apps to the Cloud through automated pipelines.
Experience includes; Linux VMs, Container Apps/App Service, Functions/Lambda, SQL Databases, Blob Storage, User/IAM, Message Queues/Buses, Managed Kubernetes, DNS, Registries, Redis, some Bastion/Firewall.

Main experience included creating new apps and deploying them in automated ways to the Cloud over various resources. Also conducted tasks such as public to Private Virtual Network resource migration and single > flex DB server upgrades.

Implemented Terraform to also create resources from templates (IaC) and managed User permissions to resources.

Scoped, researched, built and deployed various AI tools.

Backend: Support chatbot API that used a VectorDB and regular lexical lookup to intelligently respond to support queries, also tested retraining LLM models/weightings.

Frontend: 'Liveness' video login that would challenge you to move your nose into a box and would capture the 3D rotation of your face to prevent compromised facial logins using a static image (eg. someone else's photo).

Python Apps

~70% of my non-DevOps experience is Backend based, mainly in typed Python. Including; Django, FastAPI, Shopify, Pydantic, Poetry, scripting, API integration/creation, API abstraction layers, Python/framework/package upgrading, monolith re-architecting, complex bug-fixing, testing and more.

I'm a big advocate of Design Patterns and enjoy complex work in particular; I've created many different natures of solution (scripts, container apps, etc.) and worked in various major frameworks.
I've brought a lot of value to the companys I've worked for, not only within my designated role but also recognising opportunities and solutionising enhancements outside of it.

JavaScript Apps

~30% of my non-DevOps experience is Frontend based, mainly in React JavaScript, CSS and HTML. Including; React, React Native (and Expo), testing (Jest, Unit.js, Cypress), UXUI design & functionality (5+ years separate experience) and some Vue and Angular.

Also actively used Tailwind, SCSS/SASS with bem. And have converted Vue/Angular functionality into React. Have experience in implementing usual testing, and A-B test functionality through third-party services.

Tech Agnostic Experience

I've had to consume and update several other languages and frameworks commercially; Java, Ruby (on Rails), Elixir (Phoenix). Usually involving updating or adding to logic, or writing descriptive reports on long-lost production functionality which had been untouched for years.

I've also interviewed across all levels and roles; Junior to Lead devs, Frontend and Backend. And I'm particularly familiar with remote and startup environments where devs are based globally.

Other non-commercial tech experience includes; Pandas, Numpy, TensorFlow/Keras to create ML models, some game engine experience and building an enterprise-standard homeserver running on Debian Proxmox.

Grand Bequest App

Agile/Scrum in 1-week sprints - A web-app for the company Grand Bequest - users can live-capture/upload images of an abandoned building from a phone/computer and add geolocation pins to an interactive Leaflet map with extra information - the data is sent to a Firebase DB and the image is uploaded to the storage and linked to the DB entry. The company reviews the information and partners with architects/residential developers to create solutions for these abandoned buildings.

My role was full-stack, but I primarily developed all of the core back-end functionality. Built using React JS, Firebase, Leaflet, and Figma.

Logic Poker

Logic poker is a type of speed poker with an emphasis on bet sizing. Check out the github repo for an explanation on bet sizing and more about the project.

Features: Jest w/ Babel unit testing, logic for win and draw outcomes, QoL buttons and efficient UXUI, change settings.

AGAS Full-Stack App

Any Game Any Score (AGAS) is a full-stack web app which uses JAVA Spring Boot with a H2 JPA MySQL DB for the back-end and React for the front-end. It also uses the SPA PKCE Flow for OAuth2 via Okta, which passes credentials from the front-end to the back-end.

SECURITY Information is secure three-fold, firstly users must login through Okta authentication, secondly all of the back-end endpoints require a legitimate auth header to access them, thirdly the front-end URLs are secured via SecureRoute (only logged in users can view their information).

Rimi's Eyetrack

A simple but effective Python-built tkinter app for training reactions with customisable grid size, target colour, and background colour. Utilises numpy to create an invisible grid and exported with cx_Freeze.

For more features check out the github readme!

Atlas Info

Search and display key aggregated information about 100s of cities/regions! Easily check an area's safety, quality of life, education, healthcare statistics, alongside a brief bit of information to describe the area and population.

Retrieves city/region information from the Teleport API and renders them using React, chart.js, and routing. Responsive for desktop, tablet, and mobile.

Web Calculator

A custom-built calculator without using the eval() function. Utilises end-to-end testing with Cypress and features a dynamic bubble background.

Scroll through the previous answers to input them into the current calculation! Modular in function and design with grid-area, and ready to be expanded into a full scientific calculator.

RPG Dev Log

A development log for my RPG (see next slide) made with React so that people may tune in and follow the active updates! You can also view information regarding the current team and other interesting tidbits.

The main view for the Dev Log is in a blog format, check the link to view all of the game's updates and find out its current state! Or view the repo to see the React JS code.

Python RPG + Machine Learning

An in-depth RNG-based RPG with millions of weapon, armour, and creature combinations - all of which are easily generated on the fly. The working combat system is complex, featuring; randomly generated magic, a multitude of status effects, buffs/debuffs, effects over time, and more. All of the above is in the game using 1000s of lines of code and also works! Developed using GitLab and its ticket board in my free time. 2D pixel art and the music will also be developed by myself. Next stages include porting the code to the GODOT game engine.

Various NN ML projects including Regression, Classification, KNN, SVC, PCA, clustering, and other various tools within supervised and unsupervised learning.

Morse Translator

Via TDD unit testing, the Morse Code Translator was created!

Features: TDD unit testing with Jest and Babel, English to Morse, and Morse to English, Text to Speech.

Chordy Guitar

An Android, iOS, and web app made in React Native with Expo - free with ads and Paid versions exist.

Easily view chord variations for a guitar key, see the 7th chords simultaneously, or view the entire chord list! Required Python for handling file management.