Skip to content

Open to new engagements

Get in touch
MHS

Experience

12 years, 7 organisations, one consistent pattern: own the hard problem end-to-end.

The founder’s full professional history from Jun 2014 to today. Each entry lists what the system was, what was delivered and what it ran on. The PDF has the same facts in a form that fits a hiring pipeline.

Founder’s CVPDF

Jun 2023 — Present

3 yrs 4 mos

Klarna

Disputes platform & production AI systems · Remote

Payments & consumer finance

Architected and built Klarna’s next-generation disputes experience end-to-end — an evidence-first, AI-assisted system spanning mobile, web and backend services — then led a series of production AI systems on top of it.

  • Tens of millions of customers / month
  • Hundreds of thousands of AI-assisted cases
  • 15 live services migrated
What was delivered
  • Replaced a slow, agent-intensive disputes process with a system that captures and validates customer evidence at the point of dispute and builds the case presented to merchants. Serves tens of millions of customers each month.
  • Built Disputes Copilot: an AI arbitration system reasoning across customer and merchant claims, evidence, contracts, policies and business rules to produce structured, evidence-backed recommendations for human decision-makers. Operates across hundreds of thousands of cases, with continuous model evaluation in LangSmith.
  • Led delivery of an AI investigator for operational support: teams ask complex dispute questions in Slack; the system identifies the dispute, pulls live data from multiple internal systems, retrieves business knowledge and reasons over it with LLMs. Structured outputs, tool calling, guardrails and automated evals.
  • Shipped further AI tooling: a natural-language classifier that reduced disputes raised under the wrong reason, and a RAG assistant that lets customers and agents query merchant-specific returns, refund and dispute policies.
  • Owned wider disputes architecture: chargebacks, event-driven services, data infrastructure and platform work. Migrated 15 live services off legacy Jenkins pipelines, built an OpenSearch indexing service over DynamoDB dispute data, and owned observability, alerting and operational reliability.
  • TypeScript
  • Python
  • Node.js
  • React
  • React Native
  • AWS
  • DynamoDB
  • OpenSearch
  • Terraform
  • LLMs
  • RAG
  • LangSmith
  • Datadog

Feb 2023 — Jun 2023

5 mos

So-Sure

Payments platform rebuild · London

Insurtech

Brought in to overhaul an unreliable Direct Debit payment service. Reported directly to the CTO and CFO; re-architected the system as event-driven, domain-driven and serverless on AWS.

What was delivered
  • Led product design, implementation and testing of the new payment system end-to-end, resolving critical reliability issues.
  • Built on TypeScript, the Serverless Framework and AWS (SNS, SQS, EventBridge, Lambda), PostgreSQL and Docker for scalability, cost-effectiveness and fault tolerance.
  • Introduced a shift-left testing strategy across every layer of the testing pyramid and cut deployment time through CI/CD improvements in GitHub Actions.
  • At the CTO’s request, coached the engineering team on Extreme Programming principles and left comprehensive documentation for ongoing maintenance.
  • TypeScript
  • Node.js
  • Serverless
  • AWS
  • PostgreSQL
  • Docker
  • Terraform

Nov 2021 — Feb 2023

1 yr 4 mos

Cazoo

Consumer finance launch & platform engineering · London

Automotive e-commerce & consumer finance

Part of the core engineering team that designed, built and launched Cazoo’s consumer finance service on TypeScript, Node, Serverless and AWS.

What was delivered
  • Integrated Cazoo with multiple fintech and lending systems across the UK and Europe, including Santander, Black Horse and Evolution.
  • Owned the tooling and infrastructure for CI/CD using Terraform, Serverless, GitHub Actions and CircleCI.
  • Built observability and distributed tracing — error rates, request durations, cache hit/miss, mean time to recovery — with Honeycomb, Kibana, Sentry, Datadog (APM, RUM) and OpsGenie.
  • Built an ETL pipeline shipping CloudWatch and Lambda data to Elasticsearch, and internal admin tools and dashboards for customer services in Next.js and React.
  • Championed XP practice: pair programming, TDD/BDD, architecture reviews, DDD, hexagonal architecture, mutation testing, micro-frontends, trunk-based development and A/B testing. Authored ADRs, RFCs and post-mortems.
  • TypeScript
  • Node.js
  • React
  • Next.js
  • Serverless
  • AWS
  • Terraform
  • Datadog

Apr 2021 — Nov 2021

8 mos

BlokBot

Automated crypto trading platform · London

Crypto trading platform

Worked directly with the CTO to architect, build and launch an automated crypto trading platform on TypeScript, Node and the Serverless Framework.

  • 800+ customers on automated trading
What was delivered
  • Integrated with Binance over WebSocket, Lambda and EventBridge to deliver automated trading to over 800 customers.
  • Designed the system architecture, led design meetings and wrote the ADRs, RFCs, post-mortems and discovery documents.
  • Migrated legacy services from DigitalOcean to AWS microservices, and improved deployment frequency and lead time for changes with GitHub Actions and Terraform.
  • TypeScript
  • Node.js
  • Serverless
  • AWS
  • Terraform
  • PostgreSQL
  • Web3
  • Ethers.js

Jan 2019 — Apr 2021

2 yrs 4 mos

BBC

Broadcast & streaming systems · London

Public-service broadcasting & streaming

Built broadcast and streaming systems on AWS: a 24-hour online channel for BBC World 2020, the backend playout for BritBox USA, and modernised distributed publishing tooling.

What was delivered
  • Designed and deployed a full-stack 24-hour online channel for the BBC World 2020 project — React playlist editor, Node middleware on Lambda, Elasticsearch, DynamoDB and API Gateway, and a Python/AWS MediaLive playout backend — consolidating requirements from teams in London and Delhi.
  • Owned the backend playout system for BritBox USA, the streaming service from the BBC and ITV, built in Python on API Gateway, Lambda, DynamoDB, SNS, SQS, MediaLive and S3.
  • Modernised a Kafka-based publishing application (Java, Apache Camel) that lets content creators publish video to multiple playlist systems, and built the editorial portal for the BBC World Service in React.
  • Migrated a legacy Tomcat-hosted system to AWS (Lambda, EC2, auto-scaling, load balancers) with automated CI/CD in Jenkins, and mentored interns and junior developers.
  • React
  • Node.js
  • Python
  • Java
  • Apache Kafka
  • AWS MediaLive
  • AWS
  • Jenkins

Jul 2017 — Jan 2019

1 yr 7 mos

CGI

Safety-critical satellite software for ESA · London

Aerospace — European Space Agency

Worked on EGNOS, the safety-critical satellite-based augmentation system for the European Space Agency, in a team of eight modernising a codebase that had been running for over eleven years.

What was delivered
  • Developed, tested and shipped new functionality in Java and native C.
  • Wrote unit, integration and automated tests with Cantata and VxWorks to maintain 100% coverage and pass rate before deployment, as required for safety-critical software.
  • Automated testing, processes and tooling in Python.
  • Java
  • C
  • Python
  • Perl
  • VxWorks

Jun 2014 — Jul 2017

3 yrs 2 mos

Jati Ltd

E-commerce platform, built from scratch · London

E-commerce

In a team of three, built an entirely new e-commerce platform that generated £2.4 million over three years, alongside microsites, SEO and paid acquisition.

  • £2.4m revenue over 3 years
What was delivered
  • Front-end and back-end development of the e-commerce platform (HTML, CSS, JavaScript, PHP, MySQL, CDN configuration).
  • Microsite development on Magento, PrestaShop and WordPress; search engine optimisation and PPC.
  • Managed an external team of remote developers.
  • PHP
  • MySQL
  • JavaScript
  • HTML/CSS

Education

Sep 2011 — Jul 2014

BSc Information Systems

Kingston University, London · First Class Honours

Open to new engagementsOutside IR35RemoteUK-based

Have an engineering problem worth discussing?

A system that needs building properly, an AI feature that has to work in production, or a platform that keeps letting you down — tell us about it. We usually reply within one working day.

Start a conversation Email us

Elsewhere