Available for new challenges · 2026

AI products,
diagram to
production.

> AI full-stack engineer_

One person, the whole stack: the system design, the services underneath, the cloud it runs on, and the screen someone actually touches. Eight years of it — data pipelines in the UK, backend services in Japan and Ireland, platform work at AWS.

~/topology.live
pipeline — ingest to serve ingest stream model store serve skill_confidence
LLM / AI Engineering95%
Backend Systems92%
Cloud Infrastructure90%
Data Engineering88%
Platform Engineering85%
0+
Years of Experience
0+
Open Source Projects
0+
GitHub Stars
0
Of 195 Countries
End to end

Six layers, one pair of hands.

Most products get handed across five specialists and lose something at every seam. These are the layers I carry myself — from the first diagram to the thing that renders.

AI & LLM systems

RAG, agent orchestration, evals and fine-tuning. The hard part was never the first prompt working — it is the hundredth.

LLMsRAGAgentsMLOps
System design

The diagram before the code — boundaries, failure modes, what breaks at ten times the load. Most production pain is a week-one decision.

ArchitectureTrade-offsScalingReviews
Backend services

APIs built to be boring: high throughput, sub-millisecond, and still standing the morning after traffic triples.

GoJavagRPCKafka
Cloud infrastructure

AWS and GCP as code — Kubernetes, Terraform, CI/CD and observability you can actually debug at three in the morning.

AWSK8sTerraformArgoCD
Application layer

React and TypeScript wired straight through to Python and Go. The half of the product people actually judge you on.

ReactTypeScriptFastAPI
Data pipelines

Kafka, Spark, dbt and Airflow — the plumbing that decides whether your model ever sees the truth.

KafkaSparkFlinkdbt
Trajectory

The long way to full stack.

Data first, then the services, then the cloud under them, then the models on top. Nobody handed me the stack — I picked it up a layer at a time. Pick a stop.

2016 2018 2024 SCOPE NOW
2016 — 2018 Software engineer, data United Kingdom

Pipelines, ETL and warehousing on Hadoop and Spark — turning raw feeds into something a business could actually decide on. Lived in the UK and travelled Europe between deploys.

50+
Pipelines built
10TB
Processed daily
3
Countries
PythonSQLHadoopSparkETL
2018 — 2024 Backend engineer Japan → Ireland → AWS

Services and APIs across global teams: Japan first, then a relocation to Ireland and platform work at AWS. Robust beats clever when five teams depend on your endpoint.

100+
APIs shipped
99.9%
Uptime
5
Global teams
JavaSpring BootMicroservicesRESTDevOps
2024 — now Tech consultant Remote / global

Architecture and platform work for clients — and increasingly, dragging LLM systems out of the demo and into something that gets paged at 3am. The layers finally became one job.

20+
Client projects
3x
Perf improvements
31
Countries visited
ArchitectureCloud infraTeam leadershipLLM systemsStrategy
Selected Work

Four repos, four layers.

One from each layer — the infrastructure, the agents, the services, and the analysis tooling I actually run for myself. All of it public.

Open to work · 2026

Let's put something
into production.

AI product builds, senior engineering roles, and consulting where the whole stack is the job.