Leo Codron · Platform / DevOps engineer

I keep production
platforms running.

I'm a platform and DevOps engineer at CGI. I did my year in industry there in 2021 and came back full-time in 2024. Most of my time goes on Kubernetes, GitOps, CI/CD and Python automation, taking changes from the design through to production and looking after them once they're live.

3+

years in industry

3

production Kubernetes clusters I support

4

teams using tools I built

30 min

roughly saved per team in sprint planning

From commit to cluster

  1. Commit
  2. Test
  3. Build
  4. Harbor
  5. Flux sync
  6. Production

Most of my day happens somewhere along here: GitLab pipelines that test and build, images in Harbor, and Flux rolling them out to the clusters.

Experience

What I do at CGI.

DevOps / Software Engineer

CGI

Feb 2024 – present

Promoted to Level 2 in March 2026

Production Kubernetes

Support and maintain applications across three large production clusters: deployments, ingress, secrets and platform issues.

CI/CD & GitOps

GitLab pipelines with conditional jobs, tests, container builds and environment logic; Flux GitOps deployments and Harbor images.

Zero-downtime migration

Own an ongoing NiFi 1-to-2 migration for live systems, covering custom NAR compatibility, deployment design and validation.

Developer tooling

Python, GitLab and GraphQL capacity and time-tracking tools used by four teams, replacing spreadsheets in sprint planning.

APIs & testing

Tested Elastic customisation import/export tooling that coordinates multiple API calls and reshapes data.

Automation & observability

Ansible playbooks for secure VM provisioning, custom NiFi processors, site-to-site deployments and Grafana/Prometheus monitoring.

Engineering ownership

Prepare production changes, present technical designs to review boards, write investigations and review code in an Agile team.

Emerging AI

Helping try out sandboxed AI tooling in a security-sensitive environment.

Industrial placement

CGI

Jun 2021 – Aug 2022

Year in industry

Infrastructure automation

Linux infrastructure and Ansible automation for virtual machines and containers in an Agile Scrum team.

AI

Outside work I build small apps on the OpenAI API, and I use Codex and Claude Code a lot for writing, debugging and testing code. I'm particularly interested in RAG and in getting AI integrations to behave reliably and securely.

Skills

The toolbox.

AI & applications

  • OpenAI APIs
  • Codex
  • Claude Code
  • Agentic development
  • RAG & vector retrieval
  • React
  • TypeScript
  • REST
  • GraphQL

Languages & testing

  • Python
  • JavaScript
  • TypeScript
  • Bash
  • Java
  • SQL / MySQL
  • Automated tests
  • Code review

Cloud & delivery

  • Kubernetes
  • Flux GitOps
  • GitLab CI/CD
  • Docker
  • Harbor
  • Ansible
  • Linux / Red Hat
  • Ingress & secrets

Data & observability

  • Apache NiFi
  • Custom NARs
  • Elastic / ELK
  • Grafana
  • Prometheus
  • Certificates

Projects

Built on my own time.

Small things I've built for home, friends and board-game nights. Most of them use AI for the fiddly bit and plain, tested code for the rest.

Household bin tracker

BinWatch

Knows which bin goes out next, and checks the doorbell camera to see whether it actually has. A vision model reads photos from a Reolink doorbell via Home Assistant and returns strict JSON.

  • Next.js
  • OpenAI vision
  • Upstash Redis
  • Vercel Blob
  • Home Assistant

Board-game scorer

Score 7 Wonders

Photograph each player's city and it scores the game. The AI only reads the board; scoring is deterministic code over a full, tested card database covering every expansion.

  • PWA
  • AI vision
  • TypeScript
  • Tests

Party-game hub

Fun Games

Krillion, Word Race and Closest Wins for solo, 1v1 or up to ten players in a room. Krillion has 7,800+ answers and an optional Claude judge for off-list answers.

  • React
  • Node
  • WebSockets
  • Claude API

TV party game

Chameleon Board

Plays The Chameleon on the TV, controlled from your phone. Type any category and AI builds a new board, saved for next time.

  • Next.js
  • Upstash Redis
  • AI generation

Sports-court booking

Court Finder

Lists open badminton, table tennis and squash courts at two local leisure centres and books them in a tap, using the site's own JSON API. Home Assistant alerts when a slot frees up.

  • Node
  • REST
  • Caching
  • Home Assistant

Game balancing

Duel objectives simulator

A simulation and balancing system for a fan-made 7 Wonders Duel expansion, playing thousands of games to tune private objectives.

  • Simulation
  • TypeScript

Scheduled scraper

Swim timetable

Logs in every day, scrapes the local pool's Swim for Fitness sessions and publishes a mobile-friendly timetable. Read-only, with debug artefacts on failure.

  • Playwright
  • GitHub Actions
  • GitHub Pages

Home Assistant

Home automation

Lighting, presence sensing, TV controls and household devices tied together with API-driven and event-based automations, plus an AI-assisted doorbell workflow.

  • Home Assistant
  • APIs
  • Automations

University & freelance

Earlier work

A Python/OpenCV SET-card recognition algorithm for my dissertation, a Java autonomous vehicle using Dijkstra's algorithm and PID control, and commissioned Java/MySQL plugins.

  • Python
  • OpenCV
  • Java
  • MySQL

Education & training

First Class Honours

BSc Computer Science (Year in Industry)

Royal Holloway, University of London · 2019 – 2023

Dissertation on computer-vision detection and recognition of SET playing cards. Algorithms, software engineering, cybersecurity and databases.

Training

Kubernetes & Red Hat

Kubernetes administration training aligned with the CKA curriculum; Red Hat System Administration I & II.

Contact

Get in touch.