- Parallelized Machine Learning training code using Apache Spark and Arrow, speeding up overall runtime from 21 hours to 10 minutes.- Developed data ingestion framework to dynamically ingest new data sources, achieving 3x optimization on the ingestion runtime.

- Efficiently managed and processed trillions of data rows by developing and executing advanced ETL scripts using SQL and Redshift- Collaboratively designed a robust end-to-end data ingestion architecture that enabled the daily monitoring of key performance metrics on an AWS QuickSight dashboard- Streamlined metric computations by devising an automation strategy, consequently reducing the manual analysis workload for each software engineer by 2 hours a week.

- Developed ETL scripts utilizing Azure Databricks, Python, and SQL to facilitate the daily processing of 3 million rows of data into Azure Data Lake Storage- Achieved a 10x optimization of the data pipeline by employing Python and PySpark for parallel data processing on resilient distributed datasets- Engaged closely with stakeholders to convert business requirements into technical solutions, resulting in the successful and early delivery of the project by 2 weeks

- Conducted data analysis on survey results from a sample group of 30,000, providing insights to inform financial budgeting decisions.- Constructed and executed ETL pipelines using Python scripts, SQL stored procedures, and SQL Server Integration Services for data ingestion from on-premise sources.- Engineered and deployed a Data Vault model for a Data Warehouse, optimizing query performance - enabling users of the dashboard to query the database of 3 billion individuals in under a second.

Worked with the Course Director of Artificial Intelligence at Coventry University on the research of Topic Models and how it can be applied to health data.

Worked closely with the Data Science team to develop, research, and design Data Science training material and products. Attended weekly meetings to propose my ideas as well as provide update and feedback regarding current work.- Web Scraped forums with Selenium and Beautiful soup to compile a dataset of question and answer pairs. - Prepared training materials for database normalisation & statistical distributions with Jupyter Notebook and Microsoft Office.- Performed Natural Processing Analysis on a dataset of 44,000 tweets, involving pre-processing with NLTK and Topic Modelling with Latent Dirichlet Allocation (LDA) with gensim. - Co-designed internal employee benefits platform.- Prepared 2 proposals for Data Science training products.