🧪

Platform & Systems Engineer

Automation, Internal Tooling, Data Platforms

"Eliminate toil. Standardise workflows. Build durable abstractions."

architecture automation multidisciplinary builder developer tooling internal platforms

What I Do

Automate 🕷

Replace manual, fragile processes with pipelines and services.

Platform 🤯

Tooling and infrastructure that scales teams and workflows.

Abstract ♻️

Convert recurring patterns into repeatable systems.

Observe 🧪

Measure, log, and make invisible complexity visible.

Own ⚔️

Take ownership of the stack from concept to production.

Modes

solo conception → design → product architect then automate horizontal generalist own the whole stack firefighting & chaos-compatible open-source tinkerer extensibility-first design build fast, break fast, iterate

Highlights

Secure Data Platform API
Secure Data Platform API
Infrastructure diagram for secure data platform API
Annotation Platform Suite
Annotation Platform Suite
Having a cup of tea whilst my data processing library does everything
PiAudiobook
PiAudiobook
Offline, voice-controller and FOSS audiobook system powered by Python
Juggling Analytics & Practice
Juggling Analytics & Practice
Mastering juggling patterns in practice
Head Brewer
Head Brewer
Our award winning ESB Reverence from my former life as a Head Brewer
Cat Owner
Cat Owner
Frankie the cat
🔍 Click images to zoom (or unzoom)

Selected Systems

SECURITY & INFRA INTERNAL: TRUE

Secure Data Platform API (Prototype)

Problem: Fragmented workflows, inconsistent database schema and R&D teams reliant on 'snapshot' data requests.

Solution: FastAPI gateway with Workload Identity Federation (ADC), user-aware access, SQL-backed database, and Terraform-defined infrastructure. Provides R&D with data 'on-tap' for integration into workflows instead of manual snapshots.

python fastapi terraform cloud-run cloud-storage sql adc gcp docker
Outcome: Foundational platform for all company data operations; seamless, identity-based access; principle-of-least-privilege; separation of duties; reliable, integrated R&D workflows.
PLATFORMS & TOOLING INTERNAL: TRUE

Annotation Platform Suite

Problem: Existing annotation UI was slow, brittle and experiencing scaling failures. Firestore-based tabular storage hit size limits; tools violated DRY principles and were difficult to maintain.

Solution: Rebuilt the platform with a SQL backend, class-based architecture, and a highly configurable data processing library. Enabled end-to-end workflows: select datasets, process, upload and mark for annotation in 4 lines of code with no monitoring required.

python ffmpeg gstreamer sqlite docker sql multiprocessing cloud-run
Outcome: From 2 fragile tools to 30+ stable, extensible tools. Reduced annotation setup & FACS annotation time ~50%. Removed error-prone ad-hoc scripts, enabled rapid scaling and consistent deployment across studies.
PLATFORMS & TOOLING INTERNAL: TRUE

Multimodal Capture Software

Problem: Previous lab capture software was unreliable, poorly synchronised and difficult for non-technical operators.

Solution: Built a fully offline, touchscreen-driven audio/video capture system with multi-camera support. State-machine-based architecture allowed easy integration of additional sensors (e.g., heart rate monitors) and flexible configuration for new study requirements.

python pygame opencv ffmpeg hardware multiprocessing state-machine
Outcome: Supported 3-camera HD AV capture with audio/video sync; simple interface for non-technical operators; standardised and reliable collection for studies with 60+ participants; modular and extensible for future research needs.
PLATFORMS & TOOLING INTERNAL: TRUE

Datapool - Prototype Volunteering Platform

Problem: Existing volunteer sign-ups were slow, manual, error-prone, and unscalable.

Solution: Privacy-first fullstack platform for study participation with dynamic forms, targeted allocation and admin tooling.

python flask firestore firebase gcp nosql cloud-run docker
Outcome: Replaced ad-hoc Google Forms; automated participant management; location-based targeting; race-condition-safe allocations; privacy-first design.
DATA & COMPUTE INTERNAL: FALSE

JobLens - Automation Pipeline

Problem: Manual job search and triage was slow and unscalable.

Solution: Scraping, normalisation, NLP scoring, LLM summarisation and automated ranking pipeline.

python llm scraping cron sklearn openai-api json
Outcome: Fully automated job search; converted unstructured job postings into json data; reduced human workload to decision-making only.
OPEN SOURCE & SIDE PROJECTS INTERNAL: FALSE

PiAudiobook

🤯 GITHUB

Problem: Locked to proprietary audiobook ecosystems with cloud voice systems; wanted privacy and offline control.

Solution: Offline audiobook player with voice control and persistent playback.

python vosk sqlite raspberry-pi linux
Outcome: Fully offline, privacy-first, modular and portable audiobook experience.
OPEN SOURCE & SIDE PROJECTS INTERNAL: FALSE

The Zote API

🧪 GITHUB 🩸 API

Problem: Wanted a personal API to experiment and demonstrate working knowledge of Java.

Solution: Built a Spring Boot API with Infrastructure as Code (Pulumi), CI/CD pipelines, and cloud-native deployment on Google Cloud.

java maven spring-boot pulumi cloud-run cloud-sql ci/cd gcp
Outcome: RESTful API to share the grandeur of Zote the Mighty from Hollow Knight; shows Java, cloud deployment, and IaC skills; portfolio showcase.
OPEN SOURCE & SIDE PROJECTS INTERNAL: FALSE

Unified Content Platform

📎 GITHUB ♻️ WEBSITE

Problem: Content scattered across decades, platforms and mediums.

Solution: Unified platform indexing 20+ years of work with metadata, tagging, NLP recommendations and RSS feeds.

python flask firestore cloud-storage nlp RSS
Outcome: Single coherent hub for blogs, galleries, games and archives.
COMPUTATIONAL CURIOSITIES INTERNAL: FALSE

Auld Lang Syne Simulator

🖤 GITHUB 🧱 BLOG

Problem: Wanted to analyse win probabilities and difficulty of Auld Lang Syne solitaire.

Solution: Python simulator with recursive backtracking, memoisation and multiprocessing.

python multiprocessing
Outcome: Sampled 10k+ deals to compare simple vs. perfect strategies; developed scoring system for partial progress and strategy evaluation.
COMPUTATIONAL CURIOSITIES INTERNAL: FALSE

Juggling Analytics & Practice System

⚔️ BLOG

Problem: Juggling progress hard to quantify; avoided difficult patterns; wanted structured practice, progress tracking and structure archive.

Solution: Weighted random practice generator, archival system and analytics.

python pandas csv yaml
Outcome: Improved consistency in advanced patterns (552, 5-ball cascade); historical tracking & analytics.

History

Blueskeye AI

Data Ops Engineer — Nottingham, United Kingdom (January 2024 – December 2025)

  • 🡆 End-to-end data infrastructure
  • 🡆 Annotation systems
  • 🡆 Offline AV capture
  • 🡆 Automation & CI/CD
  • • Re-architected internal annotation platform; reduced tool deployment time from days to hours
  • • Migrated Firestore to PostgreSQL; enabled reliable handling of millions of annotation points
  • • Designed modular Python data processing library standardising data lifecycle
  • • Built high-performance annotation interfaces, reducing annotation time by ~50%
  • • Engineered robust offline multi-camera AV capture platform for field studies
  • • Implemented CI/CD, YAML-driven pipelines, and closed-loop automation
    • Python SQL FastAPI NiceGUI Flask PostgreSQL Cloud Storage Cloud SQL GCP Linux FFmpeg OpenCV Shell Scripting Multiprocessing CI/CD / GitHub Actions Automation Pygame

    Le Wagon

    Teaching Assistant — Remote (January 2024 – July 2025)

    • 🡆 Python & SQL mentorship
    • 🡆 Data Science bootcamp teaching
    • 🡆 Debugging & architecture guidance
  • • Delivered multiple Python course iterations to new developers
  • • Led modular architecture sessions
  • • Reinforced best practices in code quality, workflow design, and problem solving
    • Python SQL Tensorflow SKLearn Numpy Pandas

    Date & Game

    Development Team Lead — Remote (April 2021 – May 2023)

    • 🡆 Multiplayer games
    • 🡆 API integration
    • 🡆 Distributed systems
  • • Shipped six multiplayer games integrated with company API
  • • Built reusable multiplayer framework for rapid iteration
  • • Solved reconnection, resynchronisation and deterministic identity challenges
  • • Onboarded new developers and ported API integration to Unity
    • GameMaker Networking Distributed Systems Concepts Socket.io

    Self-Employed

    Indie Game Developer — Remote (May 2017 – January 2024)

    • 🡆 Independent games & interactive experiences
    • 🡆 Systems & engine architecture
    • 🡆 Full production pipeline (code, design, art, music)
  • • Developed 'Other Minds', a multi-year project with eight worlds and emergent AI gameplay
  • • Awarded Frontpage on Newgrounds with 'A Relaxing Typing Game'
  • • Narrative-driven 'Remember Mary' featured in Indiepocalypse Zine
  • • Managed programming, engine architecture, art, music and deployment
    • GameMaker Studio 2 OpenGL Systems Architecture AI Programming Narrative Design Aseprite Logic Pro X

    Matlock Brewing Company

    Head Brewer — Matlock, United Kingdom (March 2015 – January 2024)

    • 🡆 Brewing operations
    • 🡆 Product development
    • 🡆 Operational analytics & tooling
  • • Designed and brewed award-winning beers including 'Reverence' ESB
  • • Developed Excel/VBA tooling to track cask locations and contents
  • • Managed market stalls and events as brand ambassador; created Excel/VBA based scripting for till and sales management.
    • Brewing techniques Excel & VBA Event management