• Deployed a RAG chatbot using Cloud run, Elasticsearch, and Redis, to answer internal advisor questions.• Re-designed new customer credit scoring flow, improving sign-up conversion from 30% to 75%, generating millions in revenue each year. • Led the development of Heimdall, an in-house monitoring tool for ML models, hosted in Kubernetes with a MySQL back end and Streamlit front end. Heimdall supported all ML monitoring at UW and improved time-to-monitor from days to weeks for new ML models.• Overhauled batch-serving implementation which reduced Vertex AI deployment costs by 70%.• Pivoted VertexAI model serving to Cloud functions, which reduced deployment costs by 60%.• Deployed a multi-label call categorisation model using VertexAI and TensorFlow to automate the gathering of customer feedback, reducing man hours from weeks to minutes.• Presented Heimdall to audiences at Cambridge Spark, the FCA, and Mesh AI’s Data and AI Meetup.• Created a forecasting web app for the growth team which was used to inform exec-level decisions and company-wide forecasts.• Designed internal MLOps guidelines.• Created and productionised a model to identify at-risk customers which informed our communication strategies.

• Led EU item demand forecasting and stock ordering; architected a bespoke forecasting algorithm and stock ordering framework for 50+ stores across Europe.• Led the research, planning, development, and deployment in the migration to a probabilistic model for the EU item forecast.• Identified and implemented process analytics to measure and improve waste and availability decision making across all stages of the supply chain process.• (While part of Fancy) Designed and implemented a shelving algorithm which generated store planograms, improving order pick time by 18%.• (While part of Fancy) Built and owned the core ETL pipelines to transport data from Firebase to GCP. • (While part of Fancy) Created core production data tables used at the company.

• Designed a model to predict customer churn which was forecast to generate £50k+ in incremental sales on a monthly basis.• Created personalised email variants for our weekly email and A/B tested their effects - the personalised variants generated over £100k in incremental revenue each month.• Investigated the drivers of a customer’s early life journey which helped to inform the Ocado Early Life strategy.• Developed key reporting suites in Tableau to assist the Sales team and the CRM team.

• Developed an automated presentation builder to assist in Merchandising-based company-wide department reviews.• Performed a review of a change to the website’s taxonomy and fed back results and recommendations to senior stakeholders.

• Created supplier facing Tableau dashboards for Fetch which were sold to suppliers and generated a £100k+ revenue per year.