My work
Selected work across business systems, reliable AI workflows, cloud infrastructure, delivery automation, and security.
Current work
Automation Systems Engineer, AAA Accelerator
Building and supporting the booking, onboarding, CRM, payment and reporting automations a B2B coaching organisation runs on, applying software-reliability practice to a multi-system operations environment.
Revenue systems reconciliation
A demonstration that reconciles conflicting billing, CRM, community-access, booking and support records into a single reviewable customer view, with a prioritised queue for the records that need a human.
It runs on synthetic data only. The point is to show identity resolution, data-quality controls and human review working together without exposing anyone’s systems or claiming a revenue figure.
Operating platform
A private Python and SQLite platform with FastAPI services, Next.js dashboards, scheduled automations, and approval and notification flows through Slack and Telegram.
Deterministic validation wraps every LLM-assisted step: explicit approval gates, health checks, telemetry and exception handling, so a model is never the last thing between an intent and a live system.
Lead-response prototype
An interactive lead-response prototype for home-service businesses: a new inquiry triggers a voice callback, structured qualification, an availability check, and a booking or human handoff.
The web experience included consent capture, abuse controls, rate-limit states, and explicit failure paths. It is presented as historical implementation evidence, not as an active public calling service.
AI systems design
A public framework for separating deterministic controls from probabilistic model decisions in AI-assisted workflows.
The pattern puts explicit validation between model steps so an uncertain output cannot quietly become a live-system action.
Earlier DevOps work
A highly available web application deployed across Azure Kubernetes Service clusters in multiple regions, with global routing, failover, observability, and Helm-based delivery.
Earlier DevOps work
A GitOps delivery workflow for Azure Kubernetes Service using Azure DevOps and Flux, with rollback and secrets-management concerns treated as part of the deployment design.
Earlier DevOps work
A CI/CD example that adds vulnerability scanning to an AWS pipeline so application security checks happen before deployment.
Earlier DevOps work
A serverless commerce application using managed AWS services for authentication, order processing, workflow orchestration, and data storage.