Work · AI platforms, systems and leadership

Platforms, products, and the space between them.

My career has moved from product-facing engineering and open source, through startups and enterprise systems, into engineering leadership and AI platforms.

The common thread is translating between what people need, what technology can support, and what organisations can reliably operate.

Current work Public experiments

01 / NOW

AI Platforms at Bloomberg

Today, I’m a senior engineer in Bloomberg’s AI Platforms organisation. My focus is enabling employee and developer productivity, and agentic access for external clients.

I keep the details of internal systems private. The broader engineering problems are the ones I discuss publicly: developer workflows, trusted data and tools, evaluation, permissions, reliability and operational ownership.

A model may provide the capability, but those surrounding systems determine whether teams can use it safely and whether an organisation can depend on it over time.

Platform capabilities · Developer tooling · Integration · Evaluation · Reliability

Internal work remains private. The views and public projects on this site are my own.

02 / HOW I GOT HERE

Each layer changed how I see the next one.

  1. Frontend & open source

    Working close to the interface

    Samsung Mobile R&D · UI Software Engineer · 2012–2013

    Working close to the interface taught me that abstractions eventually meet real people. Technical decisions become visible through usability, performance, accessibility and the ease with which another developer can understand the system.

  2. Startups

    Prioritisation as a technical constraint

    Paxos · Lead Engineer · 2014–2016

    Startups made prioritisation immediate. Scope, speed and commercial reality could not be separated from engineering decisions, and a technically elegant solution was not necessarily the useful one.

  3. Enterprise systems

    Operational behaviour as part of the product

    Bloomberg · Software Engineer · 2017–2023

    Larger organisations introduced a different class of constraint: integration, reliability, security, institutional knowledge and clear ownership. At that scale, operational behaviour becomes part of the product.

  4. Engineering leadership

    Architecture as a coordination system

    Bloomberg · Engineering Team Lead · 2024–Present

    Leading engineers taught me that architecture is also a coordination system. It influences who can make decisions, how teams work together, how knowledge travels and what happens when a system fails.

  5. AI platforms

    Where those lessons meet

    Bloomberg · AI Platforms · 2024–Present

    AI platforms bring those lessons together. They sit between fast-moving model capabilities, product teams with concrete needs, enterprise data and systems, and organisations that require accountability and dependable operation.

03 / SELECTED EXPERIENCE

Evidence, not an inventory.

Creating leverage across teams

Context
Provisioning managed compute resources at Bloomberg used to take many months.
My role
Built a system that consolidated this into one interface.
Result
Time-to-provision dropped from months to hours, for all of Bloomberg engineering.
Trade-off
Committing to one unified interface meant a slower path to get there: early on, integrating each underlying system required shims just to get it working, then separate follow-on work to bring each one in properly as a first-class citizen rather than leaving it patched together.

Growing engineers

Context
Over 8 years, 15 early-career engineers joined and moved through my team at Bloomberg.
My role
Mentored them as part of day-to-day team leadership, not a formal program.
Result
They went on to become senior engineers and engineering managers.
Trade-off
No two people were the same — sustaining that mentorship over 8 years meant continuously adapting to very different personalities and trajectories, alongside the regular demands of shipping work, rather than following one playbook.

04 / STILL BUILDING

I still build small products outside my day job. Cal and Cine were created over one weekend with Claude Code as experiments in AI-assisted product development.

They are deliberately modest in scope, but they expose real decisions about interaction design, data integration, state, failure handling and the distance between working software and production software.

Consuming tools and abstractions directly is useful discipline. Platform decisions look different when you also have to build against them.

Weekend experiment

Cal

A visual agenda that arranges plans, places and times on one shared timeline.

Open Cal

Weekend experiment

Cine

A cinema listings board designed around comparison across time.

Open Cine

05 / OUTSIDE WORK

Conversations with small teams

Outside my day job, I occasionally have informal, pro-bono conversations with small firms exploring AI. These tend to focus on clarifying the workflow, identifying the assumption most worth testing and separating what a prototype can demonstrate from what a dependable system would require.

Get in touch