// selected work

Work

Systems I have designed, built and operated end to end — platform, cloud and AI infrastructure at production scale. Each is an anonymized case study: the problem, the approach, the outcome.

A company-wide agentic AI platform

60+ active users Sole owner of the secure, highly available AI platform 60+ staff depend on — rebuilt from a single-user open-source agent into a multi-tenant system with isolated per-user workspaces. KubernetesMulti-tenantPythonLLMSecurityHA

Multi-Host GPU Inference Pool

~120 tok/s/user (~350 agg) throughput Designed and tuned a pool of mixed GPUs across several machines that serves a stack of self-hosted models — an 80B flagship, a faster mid-size model, and embeddings — fast enough for a whole team to use at once, without paying per-seat cloud bills. vLLMGPUModel ServingLLM InfrastructureLiteLLMllama.cppEmbeddingsPerformance

Production Kubernetes Platform, Built From Scratch

7 nodes Designed and operates the 7-node, self-managed Kubernetes platform that runs the company's AI, development, and production workloads end to end. KuberneteskubeadmAmazon EKSCalicoArgoCDGitLab CI/CDRook-CephPrometheus / ThanosVeleroResticTrueNAS / ZFS

Security as a Discipline: AppSec to Estate Hardening

5+ yrs (2 titled) doing security 5+ years owning security for a RegTech group — came up through offensive security, built a repeatable pre-release security gate, and now own hardening across the whole platform as it grows. Web App PentestingOWASP ZAP / BurpNuclei / NessusVulnerability ManagementSystem HardeningShift-Left SecurityRegTech

An Automated 24/7 Incident-Response Pipeline

~$3–5k/mo cost saved Built an in-house automation that detects severity-1 incidents, finds whoever is on call, and places escalation phone calls 24/7 — replacing a paid outsourced answering service. Incident ResponseOn-CallAutomationPythonPrometheus / AlertmanagerReliabilityBuild-vs-Buy

Re-Platforming On-Prem: ESXi & Untangle → Proxmox & OPNsense

ESXi → Proxmox hypervisor Took ownership of the on-premise infrastructure and migrated it off commercial virtualisation and firewalling onto open-source Proxmox and OPNsense — the foundation every other workload now runs on. ProxmoxVMware ESXiOPNsenseZFS / CephOpen-Source MigrationWindows → LinuxNetworking

Modernising a RegTech Backend: .NET to Go

300+ endpoints ported Re-architected a flagship compliance product's backend from a stateful .NET monolith toward stateless, cloud-native Go services — cutting a key query from a 5-minute timeout to under 10 seconds and removing a hidden data-loss risk. Go.NETKubernetesArgoCDGitLab CI/CDAWS KMSPostgreSQLPerformance

AWS Estate Ownership + FinOps

~30% initial cut Primary owner of a four-account production AWS estate, where a commitment and right-sizing strategy cut cloud spend by ~30% and holds it down year after year. AWSFinOpsEKSSavings PlansCost optimisation

The Company's First BI / Data Warehouse

~600 GB → ~60 GB compression Built the company's first analytics warehouse from nothing — a ClickHouse pipeline that compressed ~600 GB of transactional data to ~60 GB and gave the business its first cross-product reporting. ClickHousePythonETLApache SupersetAirbyteKubernetes