About Me
Enterprise Backend Consultant with 20+ years of experience delivering resilient, high-throughput distributed systems across capital markets, global aviation and banking. Combining technical leadership with hands-on deliv…
Enterprise Backend Consultant with 20+ years of experience delivering resilient, high-throughput distributed systems across capital markets, global aviation and banking. Combining technical leadership with hands-on delivery: resolving complex architectural and performance challenges, modernising legacy platforms, and delivering production-grade cloud applications. Experienced across Java, Spring, Node.js, AWS, microservices and event-driven architectures.
Now extending this enterprise engineering discipline into Generative AI, building production-oriented agentic systems using LangGraph state machines, multi-model orchestration, evaluation-driven regression testing and Spring-based RAG applications. Combining deep backend engineering experience with practical AI implementation, with a focus on reliability, testability and maintainable production systems.
Experience
AI Engineer
<p>Self-directed programme building production-grade GenAI systems with enterprise discipline; all code public on GitHub.</p><p> Built an AI-assisted trading signal engine that turns an analyst’s discretionary market notes into structured, tradeable signals, using a Node.js/TypeScript pipeline on LangChain and LangGraph.</p><p> First strategy (Trident) runs intraday mean-reversion across S&P 500 instruments (ES futures, CFDs, options).</p><p> Engineered for production reliability: Promptfoo eval regression tests, safety guardrails, GitHub Actions CI, PR-Agent code reviews and Docker distribution, all built spec-first with OpenSpec, from PRD through architecture to user stories.</p><p> Orchestrated a multi-model LLM pipeline in LangGraph.js: prompt-injection guardrail and routing, then X (Twitter) sentiment search (Grok) and a Java 25 signals service (Columbo) in parallel, then DeepSeek synthesis via OpenRouter.</p><p> Selected the routing model with an eval harness: Jev (TypeSafe AI) scored 29/30 vs 25–27/30 for AWS Nova Micro with zero missed searches; low-confidence routes fall back to running both branches.</p><p> Cut synthesis cost ~45× at comparable quality by moving from Grok to DeepSeek, with per-answer cost reporting.</p><p> Streamed structured answers with a live step-trace UI over Supabase Realtime; Next.js frontend, Supabase Postgres with row-level security for per-user data, Discord OAuth via Supabase Auth.</p><p> Built a Spring Boot RAG backend covering document ingestion, semantic chunking, pgvector embeddings and similarity search, and LLM-based analysis, with cost controls, correlation tracing and prompt governance designed in.</p>
Senior Java Consultant
<p>Built the firm’s Data Mesh ingestion pipeline in Java 17 with Kafka Streams, Kafka Connect, Debezium and Apache Iceberg, giving trading teams a way to self-publish governed data products rather than queueing behind a central data team.</p><p> Reduced onboarding for new data producers to a simple formality, shortening the path to reliable, compliant data underpinning firm-wide analytics and regulatory reporting.</p><p> Ran the platform as code on Kubernetes, Helm and ArgoCD, with JUnit 5, TestContainers and Mockito gating every change through a Bitbucket/Jenkins CI/CD pipeline.</p>
Senior Java Consultant
<p>Improved performance and scalability of Cytric, Amadeus’s corporate travel and expense platform, a monolith spanning 1,000+ Oracle tables, by reworking domain-model relationships, background jobs and the Redis/Infinispan caching layers.</p><p> Migrated a legacy Servlet-based travel search microservice to Quarkus, replacing brittle in-house frameworks with a modern reactive runtime and improving startup time, memory efficiency and maintainability.</p><p> Delivered as a core engineer in a distributed Agile/Scrum team across France and Bulgaria, using Java 8/17, JUnit and Mockito, shipping through Git, Gradle and CI to OpenShift on Microsoft Azure.</p>
Lead Technical Consultant / Architect
<p>Led engineering for a Web2-friendly SaaS platform that hid blockchain complexity from non-technical users, enabling Solana-based NFT minting and wallet creation through plain web UIs and APIs.</p><p> Designed Discord-integrated backend services in Node.js and GraphQL, letting community admins reward users with NFTs from a simple command, driving real-time engagement.</p><p> Cut running costs by migrating off a third-party backend (Chainstarters) to a self-managed Heroku deployment with Postgres, Magic Link auth and IPFS (NFT.Storage), with no loss of functionality.</p>
Senior Consultant / Architect
<p>Halved Apache Spark ingestion runtime for DBS Bank through partitioning, caching and Snappy compression.</p><p> Led backend and full-stack delivery for multiple enterprise clients, including Vroom, Kinship/Whistle and DBS Bank, building cloud-native APIs, microservices and data platforms in Node.js, Go and Java on AWS and GCP.</p><p> Mentored the Vroom team by example as de facto senior engineer, championing unit testing and wiki documentation.</p><p> Architected secure, scalable backend systems such as Kinship’s Kintopia/Zootopia customer data APIs and Vroom’s account management, inventory and marketing ETL services, using Swagger, Postgres, Elasticsearch, Kafka and AWS Cognito with custom Lambda triggers.</p><p> Automated cloud infrastructure and raised the quality bar with Terraform, Docker and CI/CD pipelines, integration and load testing (Jest, Supertest, loader.io), monitoring (New Relic, Grafana) and automated validation tooling (Groovy, Unravel).</p>
Senior Full Stack Consultant
<p>Sole engineer on LSE’s course search platform: a Grails 3 / Java 8 backend, AngularJS frontend, Spring Solr faceted search and an ETL pipeline, letting students discover and filter hundreds of courses.</p><p> Delivered four weeks ahead of schedule with strong stakeholder feedback, validated end to end with Geb, Spock and Jasmine.</p>
Lead Technical Consultant
<p>Delivered 15+ complex enterprise distributed systems across Europe, Australia, the US, and Africa.</p><p> Key projects included building the national emergency alert backend (SAIP) for the French Ministry of the Interior, engineering a secure mobile banking portal for Capitec Bank (South Africa), and leading core system optimisations for the European Commission (OLAF), O2, and Amadeus Germany.</p>