Graham – Python, Typescript, AWS

This project is being carried out for a returning client, and Graham is currently executing the work to extend the functionality of the platform previously built for them.

A digital transformation initiative focused on accelerating the transition to net-zero emissions through technology enablement. The platform supported municipal and industrial stakeholders in managing climate goals, compliance, and collaboration. It included:

Graham's main scope of work includes the following:

• designed and implemented a complete CI/CD pipeline using GitHub Actions for automated build and deployment to AWS, enabling independent deployment of a newly separated codebase;
• refactored Docker build processes for both the Python backend and TypeScript frontend, isolating intellectual property within a secure new build environment;
• provisioned and configured QA testing environments in AWS to support the client’s internal testing workflows;
• conducted interviews with subject matter experts to gather requirements and document algorithm updates;
• upgraded existing algorithms as part of a broader technology modernization initiative, transitioning core logic to TypeScript.

A custom authentication middleware plugin for the Traefik reverse proxy, developed in Go. This plugin validates JSON Web Tokens (JWT) from headers, cookies, or query parameters and supports dynamic public key discovery via OpenID Connect. It features flexible claim checking with wildcard and Go template support, and configurable HTTP redirects for unauthorized or forbidden requests. Designed for secure, declarative access control in microservice and API gateway architectures.

• Designed and implemented a Traefik middleware plugin in Go to provide robust JWT-based authentication for reverse proxy and API gateway use cases. This security plugin has enterprise adoption by clients such as VW and Audi.
• Maintained 100% test coverage at all times.
• The plugin is in the top 25 most popular in the Traefik catalog.

A company sector intelligence platform developed by Brainnwave to replace outdated classification systems like SIC and NAICS with a modern, dynamic, AI-powered approach. It uses large language model embeddings to analyze how businesses describe themselves and dynamically cluster firms into evolving economic sectors. This enables more accurate mapping of sectoral shifts, identification of emerging industries, and timely analysis of economic trends. Mosaic provides sectoral benchmarking, economic profiling, and growth diagnostic and forecasting capabilities for policy analysts, economic developers, and research institutions.

• Conceived the product and developed the initial proof of concept for semantic clustering technology in Python during an intensive week-long pair programming sprint with the lead data scientist.
• Oversaw all aspects of development across frontend, backend, data engineering, and data science. Cultivated a culture of engineering excellence by establishing clear coding standards, championing test-driven development (TDD), enforcing modular service design, and leading rigorous code review practices to ensure correctness, flexibility, and long-term maintainability.
• Led the end-to-end design and implementation of the platform architecture, building scalable, cloud-native infrastructure for production and test environments.
• Designed the security model and implemented a secure, JWT-based authentication and authorization system using Python, TypeScript, Go, Traefik, Auth0, and OpenID Connect.
• Developed a robust CI/CD pipeline enabling deployment from Git -> QA -> client production environments, supporting rapid iteration, fully automated adherence to quality standards, automated testing, and seamless live rollout.
• Directed technical validation processes to assess model interpretability, clustering coherence, and semantic accuracy.
• Led technical engagement with partners and clients to capture requirements and translate them into actionable product features and roadmap priorities.

Brainnwave is an award-winning data and AI consultancy specializing in the design and delivery of bespoke decision intelligence platforms. Founded to bridge the gap between complex data and real-world decision-making, Brainnwave integrates advanced analytics, machine learning, and scalable cloud infrastructure to help organizations transform raw data into actionable insight.

• Co-founded and scaled Brainnwave into a leader in AI-powered decision intelligence.
• Served as a hands-on technical leader in a player/coach role, actively contributing code while mentoring team members and guiding architectural and engineering decisions.
• Architected and led the development of an award-winning decision intelligence platform, powering SaaS solutions for 30+ enterprise clients across multiple industries and problem domains.
• Led all day-to-day aspects of software development, DevOps, database, and data science workflows, with daily hands-on development of the most critical components in Python, Go, and TypeScript.
• Built high-performance engineering teams: recruited and mentored exceptional technical talent, fostering a culture of excellence, autonomy, and responsibility; maintained a 4.8 Glassdoor score over nine years.
• Directed a £2 million annual technology budget, ensuring optimal resource allocation for scalable, high-impact solutions.
• Enabled enterprise growth: secured and delivered on twelve FTSE 250 and S&P 400 level clients, generating multi-million-pound revenue through strategic client engagements.

A Python-based, AI-powered distributed web scraping and content analysis system developed by Brainnwave to extract meaningful business information from company websites. It automatically crawls company domains, identifies relevant content such as “About Us”, “What we do” and “Our Services” sections, and stores extracted data in a pluggable database backend for downstream processing. The system integrates natural language processing from spaCy and NLTK to filter, match, and prioritize valuable textual content. It dynamically switches to headless browsing for JavaScript-rendered pages using Selenium, and operates in a scalable containerized architecture.

• Led the design and implementation of a distributed web scraping system using Python, Docker, and PostgreSQL.
• Developed intelligent content extraction workflows using BeautifulSoup, NLTK and spaCy, including a custom NLP-based semantic matcher to locate business-relevant sections of web pages.
• Engineered a hybrid parallelism model combining containerized horizontal deployment with in-process thread pooling, orchestrated via a PostgreSQL-backed task queue, enabling massively parallel processing of websites with efficient resource utilization across distributed environments.
• Integrated Selenium and Chrome WebDriver to support dynamic content extraction from JavaScript-heavy sites.
• Built robust filtering logic to exclude parked domains, media links, and irrelevant content, improving efficiency and data quality.
• Integrated pluggable proxy services to gracefully handle geoblocking, rate limiting, and bot detection mechanisms, enabling reliable and resilient scraping of geographically restricted or protected websites.
• Collaborated with data science teams to ensure extracted content into downstream semantic classification systems produced optimum results.

A real-time market intelligence and sales enablement platform developed by Brainnwave for Aggreko, a global provider of mobile and modular power solutions. AMI integrates internal CRM data with external sources such as Bloomberg, Platts, and satellite imagery to deliver a unified view of global commercial opportunities. The platform features geospatial analytics, ML-driven lead scoring, customer segmentation, and sales forecasting, all accessible through a no-code web interface. Deployed in under three months, AMI significantly reduced sales cycle times and drove measurable revenue growth.

• Led the architecture and development of a scalable, real-time analytics platform integrating diverse data sources, including financial data, satellite imagery, geospatial vector and raster data, and CRM systems.
• Built interactive geospatial visualizations to support opportunity discovery, prioritization, and market segmentation.
• Developed machine learning pipelines for customer segmentation, opportunity ranking, and predictive sales forecasting.
• Integrated real-time data feeds from sources such as Bloomberg, Platts, and S&P.
• Delivered a production-ready MVP in under 12 weeks, supporting global rollout and user onboarding.
• Achieved tangible business impact: 25% sales pipeline increase, $8M incremental annual revenue from previously unknown opportunities, and over 1000% ROI.
• Designed a modular system architecture enabling rapid feature delivery and long-term adaptability across multiple use cases. After the initial solution for Aggreko, the platform evolved to power SaaS solutions for 30+ further enterprise clients across multiple industries and problem domains, including energy, social housing, food and beverage, engineering, transportation, and urban development.

PeopleBrowsr is a social analytics and data mining company that developed one of the earliest large-scale platforms for analyzing and visualizing social media networks. The company provides tools and APIs to track influence, sentiment, and engagement across platforms like Twitter and Facebook. Its flagship products have included Kred, a social scoring system for measuring influence and outreach, and SocialOS, a platform-as-a-service (PaaS) for building social networks using pre-built modular components. PeopleBrowsr is known for managing and indexing the full Twitter Firehose for deep analytics and has worked with enterprise, government, and academic clients to deliver real-time social intelligence at scale.

• Designed and implemented scalable, high-throughput distributed search and analytics engines in Python, enabling real-time querying of a two-petabyte trillion-message distributed database across 20+ nodes.
• Designed and implemented a relations graph analysis database for a DARPA research project. The database system stores all publicly accessible relations information from Twitter and Facebook and provides cluster analysis, champion and influencer detection, and shortest paths between nodes.
• Developed a distributed Twitter Firehose receiver that can process over 10,000 messages a second from Twitter, annotate them with multiple metrics and indicators, and index each tweet in real-time.

This was a hands-on technical leadership role carrying responsibility for the safe operation of more than 200 of Bloomberg’s mission-critical market feeds. Written in C++, the Bloomberg Market Data Feed framework provides platform abstraction, threading, messaging, and core infrastructure for market data and trading interfaces. In continuous use since 2003 to build and run all feed-handling systems within Bloomberg globally. Routes over 16 billion market events per day with peaks of 1.8 million messages per second.

• Managed the two teams responsible for all European exchange feeds, news feeds, and trading interface systems;
• Envisaged, designed, and led the development of the Bloomberg Market Data Feed Framework.
• Improved time to market by more than 3x for new exchange feeds.
• Reduced exchange feed failures by more than 7x since introduction.
• Enabled Python to be used for the first time with Bloomberg feeds.
• Provides exchange interface for Bloomberg Sell-Side Execution and Order Management Solutions (SSEOMS).

Acxiom is the recognised global leader in real-time, multichannel customer data integration and analytics, and is the world’s largest processor of consumer information.

• Set up Acxiom Australia as Chief Architect and one of the company’s four founding employees
• Was responsible for the specification, design, and hands-on implementation of the company’s entire technical infrastructure, including hardware, software, and a two-terabyte data warehouse – then Australia’s largest consumer database.
• Grew Acxiom Australia from four to 160 employees in two years, driving annual revenue to A$14M through hands-on technical leadership.
• Designed and implemented an emergent-AI-based parsing engine in C++ for Acxiom’s flagship product, AbiliTec, enabling global deployment into all English, Spanish, and German-speaking markets.
• Designed and presented a successful RFP bid to Vodafone, securing a £20M, five-year contract for their“Single Customer View” data warehouse. Led technical implementation from requirements capture, through development, to user acceptance testing.