MP3 Home Player
This was the first time I brought hardware, server software and a desktop application together as one product: a standalone home audio unit with its own networked music library.
A builder’s story
How the builds connect.
Looking back, the work feels less like a list of jobs and more like one long conversation with the same kinds of problems. Each build taught me something, and many of those lessons found their way into the next.
Eight overlapping chapters. Forty-seven builds, startups and ideas. One route through them.
Build
Startup
Idea
Chapter 01 -
Useful things become products
I started with small, practical things I wanted to fix. Before long, that curiosity had grown into hardware, publishing, digital services and my first attempts at building companies.
A self-taught Visual Basic tool led me towards a technical degree, connected products and, eventually, professional financial software.
- Car & Motorcycle ManagerI built a simple way to keep on top of servicing, repairs and running costs.
- MPFrequency.comNew MP3 players were hard to compare, so I built a place for independent reviews.
- MPFrequency.comThe review site became my first attempt to turn useful content into a business.
- Cool Oak TechnologiesA small company built around practical web, server and bespoke software work.
- LiquiPakA different kind of problem: what if the service came to the crowd instead of everyone queuing?
Why these projects belong together
Looking back, I was already doing the thing I still enjoy most: noticing an everyday frustration and wondering whether a thoughtfully designed system could make it easier.
Then the setting changed completely: from personal and consumer products to the dense operational world of hedge-fund trading.
Chapter 02 -
Learning the trading workflow
I entered financial technology through the practical end of the problem: first capture a trade, then bring many trades together, then understand the whole lifecycle around them.
At Beauchamp Financial Technology, browser interfaces, Oracle data and C++ systems turned my product curiosity into real financial-market craft.
TradeManager
TradeManager brought everything I had been learning into one enterprise product, from rapid trade capture and electronic execution to the workflow surrounding both.
- Web-Based Trade Capture for Hedge FundsMy first step into trading software replaced fragmented processes with a clearer browser workflow.
- BetSweeperA side experiment in making sense of fragmented exchange data and acting on what it revealed.
- Point-of-Sale Betting TerminalsI explored a venue-based route into exchange-style markets, then decided not to pursue it.
- Web-Based Trade Blotter for Hedge FundsThe capture screen grew into ordering, execution, allocation and the rules around them.
Why these projects belong together
With each build I understood a little more of the job around the screen. What began as recording a trade became a consolidated view, then a product responsible for the wider workflow. I would return to that pattern years later.
Once I could see the workflow as a whole, the next question felt natural: could software also improve the decisions moving through it?
Chapter 03 -
Information becomes trading edge
Inside a hedge fund, I saw how much better the work became when research, communication, market data, portfolios and analytics stopped living in separate tools.
The software was no longer simply supporting the front office. Built closely with the people using it, it became part of the fund’s recognised advantage.
Equity Research Platform for Hedge Fund
This platform connected the full decision: the research behind an idea, the people responsible for it, its place in the portfolio and risk, and the trading that followed.
- ICQ Trade ChatI used ticker rooms and keyword broadcasts to get relevant research to the right people sooner.
- Ticker NotificationRecommendations, alerts and confirmations reached people through channels they could shape themselves.
- Option Price Chain Lookup ToolA dense Reuters data set became a view that people could actually use to think.
- Portfolio Rebalancing ToolI turned the knock-on work from subscriptions and redemptions into clear, structured rebalance orders.
- TraderAnalytics.comTrading behaviour and profitability became useful feedback for coaching, not just another report.
Why these projects belong together
They all try to shorten the distance between knowing something and doing something useful with it. Research reaches the right person, data becomes understandable, portfolio changes become orders and performance becomes a chance to learn.
By then, the tools were starting to feel like products in their own right. TraderAnalytics became the starting point for Capital Coders.
Chapter 04 -
Founder tools and operational scale
This chapter taught me two different kinds of scale: how to turn specialist software into a company, and how to make complex financial operations dependable across countries and teams.
Capital Coders opened the founder path. BTG Pactual widened my responsibility from individual applications to international development capability.
Capital Coders Limited
TraderAnalytics grew into a London hedge-fund software consultancy. Suddenly I was balancing product ideas, client delivery, financial-domain engineering and the day-to-day reality of running a company.
- Fund Risk Management and Hedging CalculatorA practical way to turn portfolio exposure into the trades needed to offset it.
- ClickNowPayLater.comI wondered whether people could spread a cost without being pulled out of the purchase.
- PostboxA deliberately calm idea: turn chosen digital updates into a small physical morning briefing.
- DB2EmailMore than 50 reports stopped being a manual chore and arrived automatically.
- Trade ReconciliationBroker and internal records were brought together so breaks and missing trades could not hide.
- “Code Art”A playful thought: source code can be a visual object as well as something functional.
- High-Frequency Trading Monitor ToolkitWhen software moves at market speed, people need clear oversight, health controls and a reliable off switch.
Why these projects belong together
Running a company and supporting global operations taught me the same lesson from different directions: important work cannot depend on one person remembering everything. These tools made specialist knowledge repeatable and visible.
Career-wide operational footprint
A broader view of places where I worked across roles and years, beyond the date range of this chapter.
Global footprint
Built across markets and time zones.
Hands-on delivery and technology leadership across Europe, the Americas and Asia.
- EuropeLondonUnited Kingdom
- EuropeGenevaSwitzerland
- AmericasNew YorkUnited States
- AmericasSão PauloBrazil
- AmericasRio de JaneiroBrazil
- AsiaSingaporeSingapore
- AsiaShanghaiChina
Automation was no longer just helping people trade. Soon, the software and the trading strategy would become the same system.
Chapter 05 -
Code enters the market
This was a turning point. The code was no longer beside the investment process; it was making portfolio decisions. At the same time, my ideas began reaching beyond markets into personal data, trust, place and organisational life.
My role moved from builder to systematic trader, then into international delivery and global technology leadership.
Systematic Trading Strategy
I co-designed and back-tested a multi-portfolio, multi-asset model, then automated the journey from signal to execution. It returned 10% before being discontinued in 2012.
- Syscretionary.comA question I kept returning to: where should human judgement end and rules begin?
- API of YouWhat would it look like if people truly controlled their identity, content, consent and value?
- Skiverts.comA playful attempt to see an overlooked physical surface as a new resort media channel.
- EstateBox.comA trusted home for the records, wishes and instructions that families should not have to hunt for.
- OrbitThis.comAn idea for leaving personal memories and local knowledge in the places where they mean something.
- Capability MatrixI needed a clearer way to see, compare and govern a complicated delivery programme.
- TechnologyManager.comThe same instinct applied to an IT estate: make ownership, suppliers, assets and spend easier to see.
Why these projects belong together
This is where my field of view widened. I was still building systems, but the questions now included who owns personal data, how context survives and how people make sense of complicated organisations.
By then, architecture, markets and product thinking no longer felt like separate parts of my experience. They were ready to come together in a modern SaaS company.
Chapter 06 -
Trading systems become SaaS products
Years of building one-off institutional systems came together as a connected product family: order management, communication, notifications and knowledge sharing designed side by side.
With HEDGD, I was writing code while also carrying responsibility for the company, the product and the architecture behind a multi-asset workflow platform.
HEDGD Limited
HEDGD was my attempt to bring the parts together as a company: multi-asset order management, structured execution, useful notifications and better ways to share knowledge inside a firm.
- Web-Based Order Management SystemA modern multi-asset SaaS platform built around the real detail of allocation, charges and execution.
- Knowledge Matching AppI wondered how much useful knowledge was already sitting quietly inside trusted contact networks.
- Order ChatTrading instructions could begin as a conversation without disappearing inside one.
- FinTiqA calmer way to search, pause and organise updates that otherwise move too quickly.
- MicroReportsShort, focused reports made useful knowledge easier to publish, follow and discover.
Why these projects belong together
I kept running into the same human problem: important things arrive mixed in with everything else. These products gave instructions a workflow, updates a controllable channel and short reports a place where others could find them.
Harkster found a customer with a clearer macro-research problem, and the work moved toward the flow of institutional information.
Chapter 07 -present
Research becomes AI-assisted
By now, several old questions had met in one place. How do you keep useful information? How do you discover what matters? And how do you give busy people enough context to act with confidence?
My current work brings financial research, architecture, product judgement and hands-on engineering together. It does not replace the earlier chapters; it depends on them.
Harkster
Harkster uses AI to help institutional researchers get more from the information they already receive: finding consensus, overlooked trade ideas and market-moving events through tailored assistants and integrations.
- LinkBoxI was tired of useful links disappearing into conversations, so I imagined a private, searchable home for them.
- Harkster LimitedThe company brings my institutional research and market-intelligence products under one roof.
- ‘Tweet Deck’ for Macro ResearchScattered publications, podcasts and news became one research workspace that people could shape for themselves.
- JotJarA nearby branch of the journey: collect feedback simply, then make sure someone responds.
- Non-Slip Burger BunsNot everything has to become software. Sometimes a small physical redesign is enough.
- DiscoverThis.comA simple belief that still matters to me: recommendations feel better when they come from people we trust.
Why these projects belong together
The technology is new, but the frustration is familiar. Valuable information still arrives fragmented, quickly and without enough context. AI now helps, but it sits inside a workflow shaped by years of building research and market systems.
The questions underneath this work are still moving, even as the tools begin to change again.
Chapter 08 — ? –present
The tools are changing again
Software is becoming easier and cheaper to make. Interfaces may become conversations, agents or something else. That changes the questions around a product without settling what is worth building.
I know something important is changing. I don't yet know where this chapter ends.
Questions this chapter leaves open:
- When code becomes cheaper to produce, which problems still deserve a product?
- Will interfaces become conversations or agents, and how might that change the form of software?
- How should knowledge be structured when people and agents both need to use it?
- Where does deep knowledge of a domain or process matter most?
For now, the question mark stays.
What kept returning
Six threads I kept following
I did not name these themes at the time. They are simply the patterns I can see now when I step back and look at the work as a whole.
Workflow
I keep wanting to understand the work around the screen, then remove the friction that gets in people’s way.
Research & discovery
From independent reviews to AI-assisted market intelligence, I have always been drawn to helping people find what matters.
Communication & attention
The right information only helps when it reaches the right person, in time, without adding more noise.
Markets, risk & automation
Markets taught me to respect speed, uncertainty and the importance of keeping people in control of complex systems.
Trusted personal information
Many of the ideas ask the same human question: who should own our information, memories, context and recommendations?
Turning tools into companies
MPFrequency, Capital Coders, HEDGD and Harkster each began with the hope that a useful specialist tool could become something durable.
The unfinished present
This chapter isn't finished.
Harkster remains part of the story. Chapter 07 and this newer chapter overlap; the questions about what comes next are still open.