Amazon
London, United Kingdom
Worked within Amazon's EU last mile capacity planning analytics team supporting on AI and automation initiatives. Developed a full-stack analytics application supporting time critical cross-team and cross-organisational processes related to planning replacing a legacy Excel and Power Bi based process.
Worked in the analytics team of the EU capacity planning team. Architected and implemented a full stack analytics platform supporting daily planning processes across cross-team and cross-organisational processes. Carried out query optimisations on the data pipeline helping to cut the time critical pipeline runtime from 41 minutes to 3.5 minutes. Developed an accompanying MCP server to facilitate the automation of the process through user defined business logics across the dataset.
Learnt development practices on AWS including Lambda functions, Athena databases and queries, Redshift tables and SQL, AWS Glue, S3 buckets, EventBridge, IAM permissions management systems and DynamoDB tables. Understood optimisation techniques on SQL from ground including order of execution optimisations, cadence based query scheduling helping to decrease core query pre-existing query execution speed by 91%. Worked with Athena based datalakes, Redshifts and DynamoDB tables understanding the best ways to structure data based on usage patterns and business requirements.
Developed strong communication and organisational skills by working with stakeholders spanning different teams and organisations handling obtaining and executing wide ranging business requirements. Whilst also presenting to both technical and non-technical audiences across the teams helping gain alignment on underlying on legacy issues with the previous data pipelines to create and update the pipeline with new metrics aligning to the modern business requirements. Developed my project management skills to deliver the initial product set before key business events within the short timeline enabling the wider user adoption which feed further feedback later in the placement. Whilst also focusing on the core feature set required and critically solving for the key issues first to then allow later non-critical expansions of the tool to be enabled.