Raul Landa

In practical terms, my work helped Fastly improve fleet efficiency by co-optimizing infrastructure capacity and service performance. To achieve this I coded machine learning algorithms in Go, Python and jinja-driven multistage SQL pipelines running on Bigquery and orchestrated using dynamically generated DAGs using Google Cloud composer (Airflow). I also focused on demistifying ML within devs.

My work in Sky revolved around creating reusable components to automate decision making and plan execution for network capacity planning. I created a traffic-matrix forecasting component that fed a what-if network simulator that analysed multiple failure scenarios; the results were used to plan for CDN and peering/transit capacity. I also helped model how video impairments harm user engagement.

My research encompassed novel Internet protocols and mathematical models, aided by purposefully- built network simulation, measurement and analysis software. I helped tutor MSc and PhD students, define research strategy and delivery of research outcomes, wrote research proposals, and collaborated with industry partners.

My work allowed large telecom companies to better understand their network performance by developing and integrating network failure and performance management software components in Java and Perl.