I am a software engineer (SWE) working in the ML Infra team.

I am building software and optimising open source & in-house models to reduce model cost per query and to improve serving latency for GenAI inference workloads

@ Founder's Office

Worked on AI integration with Trading212 app features. I reported directly to the co-founder.

I was a software engineer in Large Models SRE team (in ML Infra serving).

We were responsible of managing shared use of high-demand Google AI models (e.g. Gemini, Veo).

I led the design, implementation and delivery of a billing solution for TPU and GPU usage for popular models like Gemini and Veo.

* It was serving multiple internal customers; a 1.5-year cross-functional project involving 3+ teams.

* I owned system architecture and core implementation end-to-end.

* I carried out processing of large volumes of structured logs and time-series metrics for cost attribution.

I contributed to

* Capacity management of model servers and auto-scaling

* Client isolation and fair-use components for shared AI model infrastructure.

* Oncall duty for Large Models

Developed trading systems operating in football betting exchanges (i.e. Betfair)

* Developed order creation engine to do market making

* Applied portfolio management techniques to optimize exposure and P&L, reducing variance and risk.

* Identified a critical live match data evaluation flaw, which was estimated to cause £500k loss.

* Used Python (with pandas & NumPy), SQL and Kubernetes.