- System design and implementation ownership of two projects (live ML inference service for ETA predictions & internal data service for pre-caching analytics data using Cube)- Developed an event-based Java service for the aforementioned Eta prediction tool- Neptune experiment tracking tool adoption implementation in ML team- ML Pipeline setup for model training and deployment- Led ML model hosting & training migration from AWS Sagemaker to Google Vertex AI- Numerous data analysis efforts understanding quality and information extraction of multiple messy datasets in shipping industry- Setup code quality checks in ML projects (black/pylint/test coverage)- Model prediction quality tracking dashboard setup in Looker, maintenance and dissemination in company

Part of Beacon AI team - let's see what happens

• Set up Kubeflow Pipelines in Azure for the ML team to use large compute resources efficiently for traffic prediction training and predictions, and setup an environment for continuous improvement.• Helping the ML team transition away from unrepeatable and opaque development practices to more productive and collaboration-friendly methods.• As part of a data analysis piece, and given public bus movement data, developed a method of identifying traffic pinch-points on a road network using Jupyter notebooks and Kepler.• Replaced and simplified the entire monitoring and logging system used by the company's customer platform, consolidating to a single, reliable view.• Reduced ongoing cloud infrastructure costs by 25% through large-data sorting and Kubernetes cluster resource management.• Overseeing security measures for secrets management, data ingestion framework and database rescaling/migrating data to an appropriate platform for analytical data work.• Setup all infrastructure, CI/CD processes and initial codebase for our move to using Dagster as a principal data pipelining tool, to simplify current and future data connector development and improve resilience and visibility.• Imported all manually defined infrastructure definitions into Terraform and standardise usage of AWS and Kubernetes resources for easy future developments.• Implemented new continuous integration pipelines and more user-friendly scripts for Helm deployments.• Setup the infrastructure and initial pipeline for future ML applications deployments, using a combination of Dagster, Ray and MLFlow.

• Implemented and deployed a new optimisation platform based on Kubeflow for internal teams to use - turned a manual process of spinning up instances, and loose use of GCP into a simple serverless command line interface experience.• Key member of the team during our creation of a demo application; managing progress, unblocking the team whenever needed; contributing code in frontend, backend and platform side; and keeping everyone pointing in the same direction• Kick off productionisation of a workforce optimisation offering: planning architecture of systems and refactoring demo work into a production grade system.

• Developing our data science processes by testing out Kubeflow• Deploying Kubeflow and using as platform for one of our client projects. All successful until coronavirus hit and momentum in the platform wained and I was moved to a different project.• Maintaining our internal batch processing system

• Building a secure, production-grade in-house batch processing application in Go, deployed using terraform• Set up all CI systems

• Building ETLs as new insurance providers joined our company platform (Talend)• Building custom integrations into systems such as weather data and Elasticsearch APIs to enrich our available data• Providing support on our analytics platform (Yellowfin)• Set up all our ETL systems when we expanded to supporting insurance providers in Australia

On rotation around the IT dept, working mostly in Java and Groovy. 3 months on the test team, creating automated test and improving the web-based in-house test framework being used here. 6 months on the underwriting team, providing developer support for both internal and client underwriters. my roles mainly involved extracting information from and altering underwriting rule sets on request. This contributed to the company library of reusable groovy scripts to improve rulesets. Then on the MI team, working mostly in data integration. In the process of finishing my second etl job, I have created one job to consolidate and make available new build performance data, and another for funnelling and making available server status history for company hosted services, on our bi platform. I have also been involved in creating a number of reports, analysing a multitude of trends including third party agent selling behaviours. Soon I will begin work on improvements to our software involving Bayesian networks to improve decision making.

In charge of redesigning the union’s websites and email template, as well as overseeing a number of events run by the RCSU. The websites include the sciencechallenge.org and the rcsu.org/exec websites.

handling all finances including product sales, yearly budgeting and claims handling. As well as this, also assuming role as vice president of the society, taking responsibility for much of the organisation surrounding DJ events, meetings and miscellaneous administration with the Imperial College Union.

Regularly proving the music for social event at the Imperial College Union club. Each time I have been able to provide a reliable service, adapting to the crowd's response to tailor music selection to them. I have also been able to DJ at venues around London for other private events, as well as Imperial College's annual Summer Ball, alongside famous acts such as Zane Lowe and Pendulum.

Overseeing the release of new features on the trialreach website and assuring that the product is of the highest standard, providing a user friendly and intuitive experience for customers. Addition, I am in charge of developing robust and structured testing methods for the key aspects of the site. During my time I also became involved in many other aspects of the company, including web development, providing technical support for marketing as well as being involved in user experience testing of the website.