Theo Windebank
London, UK · theowindebank.co.uk · github/twindebank · linkedin/theowindebank
Experience
Gradient Labs
Staff Engineer
Jul 2024 – Present · London, UK
- Most tenured engineer and one of the most prolific - #1 author of the surviving codebase, with contributions spanning backend, infrastructure, frontend, and tooling.
- Built MCP infrastructure end-to-end: connectors allowing companies to expose their own tools to the agent via MCP, and servers exposing Gradient’s APIs to customers - including OAuth flows, connection state management, and frontend management UI.
- Built the resources system (how customers deterministically provide data to the agent), the tools system (how the agent fetches data and interacts with customer systems on demand), and the foundations for human-in-the-loop flows - including general asynchronous tool execution.
- Significant contributor to outbound conversations (that’s me in the video!); designed and built the batch system enabling the agent to initiate and manage large volumes of customer conversations at scale.
- Built real-time streaming infrastructure for the conversation timeline, giving full visibility into agent reasoning, tool calls, and knowledge retrieval as they happen.
- Major contributor to third-party support platform integrations: Freshdesk, Freshchat, Zendesk, and Intercom.
- Thrives on ambiguity - prefers a problem statement over a spec, and takes ownership of the full solution: shaping the design, pulling in the right people (design, frontend, data, other engineers), and driving it through to delivery.
- Extensive Temporal experience - used throughout the platform for durable execution: async tool calls, resource refresh pipelines, outbound batch orchestration, conversation lifecycle management, and more.
- Owned infrastructure across the board: Terraform, Grafana/Prometheus monitoring, alerting, and CI/CD - while regularly shipping product changes as a fully cross-stack engineer.
- Early Claude Code adopter; built and advocated for worktree-based tooling that let the team parallelise AI-assisted development and meaningfully increase velocity.
Monzo Bank
Senior Backend Engineer
May 2024 – Jul 2024 · London, UK
Backend Engineer
Jun 2022 – May 2024 · London, UK
- Founding backend engineer on the squad that rebuilt Monzo’s Home screen - one of the most central surfaces in the app, serving both customers and internal teams building on top of it.
- Collaborated with backend and mobile engineers to build the ‘Overview’ framework: a server-driven, extensible system allowing other squads to contribute to and customise Home, effectively making the squad a platform team within Monzo.
- Became deeply fluent in Go, taking on complex backend work across a high-traffic, cross-team surface.
Senior Data Engineer, Analytics
Aug 2020 – Jun 2022 · London, UK
- Key contributor to Monzo’s data stack - a centralised BigQuery/dbt/Airflow platform serving ~150 data contributors across the company.
- Designed and built an extension framework for dbt that replaced a hard-to-maintain internal fork, enabling faster upgrades and modular customisation across ~6,900 models.
- Led data governance work including access control patterns and tooling for managing sensitive data at scale.
Data Engineer, Analytics
Apr 2019 – Aug 2020 · London, UK
- Rolled out dbt across the company; designed supporting infrastructure and onboarding docs used by over 100 engineers.
- Designed and led development of a distributed access control system written in Go for sensitive user data.
BBC
Interim Lead Data Engineer
Dec 2018 – Mar 2019 · London, UK
- Assumed lead engineer role during a 4-month rotation while the team restructured from one team of 15 into two smaller squads, introduced OKRs, and pivoted to new objectives.
- Led recruitment of a principal engineer: wrote interview questions, evaluated candidates, and assessed team fit.
Data Engineer
Nov 2017 – Mar 2019 · London, UK
- Part of BBC Datalab from its early formation - helped build BBC+, the first BBC app to serve fully automated ML recommendations, built from scratch on a Kubernetes/GCP microservice architecture.
- Worked on a subsequent internal ML platform, including deploying JupyterHub with audited access to user and content data, and building tooling to version data and graph pipeline dependencies for full reproducibility.
- Spoke at PyData London 2018.
Cue Sense Ltd.
Software Engineer
Jun 2017 – Nov 2017 · London, UK
- Built software to help visually impaired people interpret non-verbal cues using computer vision and machine learning.
Oxford Machine Learning Research Group
Software Developer Intern
Jul 2016 – Sep 2016 · Oxford, UK
- Developed ML techniques to detect mosquito presence, species, and gender from audio recordings; built upon in my fourth year project, graded 83.
- Built a fully-featured Python package as a test-bed for detection algorithms, aimed at open source release.
- Ranked in the top 5% of students who have worked with the Oxford-Man Institute.
Education
University of Oxford - MEng Engineering Science
Sep 2013 – Jun 2017 · St. Catherine’s College · 2.i (68)
Interests
Running
Hackney Half Sep ‘21 · Royal Parks Half Oct ‘24 · Kingston Half Oct ‘24 · TTP Cambridge Half Mar ‘25 · Brighton Marathon Apr ‘25 · Oxford Half Oct ‘25 · TCS Amsterdam Marathon Oct ‘25 · EDP Lisbon Half Mar ‘26 · Camino Lea Valley 50km Ultra Mar ‘26
Tour Aotearoa
Cycled the length of New Zealand - a 3,000km bikepacking route from Cape Reinga to Bluff - self-supported and camping throughout, completed in 5 weeks, Jan–Feb 2024.
Smart home
Local Home Assistant deployment; interested in home automation and keeping things off the cloud.
Skills
Go · Python · MCP · Claude Code · Temporal · Distributed Systems · Kubernetes · Terraform · GCP · Docker · Grafana · Prometheus · Kafka · SQL · dbt · Airflow · BigQuery · ElasticSearch · TypeScript / React / Next.js (Claude Code assisted) · Machine Learning · Encore · Public Speaking