AhmedIrfan

End-to-end ML lifecycle on enterprise AWS.

• Managing the end-to-end machine learning lifecycle on an enterprise AWS platform, empowering 50+ data scientists to seamlessly transition from project onboarding and Kubeflow environment provisioning to full scale model training.
• Developing a Gen AI chatbot leveraging Retrieval-Augmented Generation (RAG) against comprehensive ATO knowledge bases, leveraging LangChain, AWS etc, automating responses to 10,000+ monthly queries and reducing manual query resolution time by 40%.
• Performing critical BAU operations and ML infrastructure maintenance, executing monthly Amazon EKS cluster patching and optimizing model inference pipelines via KServe and ModelServe to guarantee 99.9% platform uptime.
• Administrating centralized ML libraries and dependencies via Artifactory, ensuring secure, standardized package distribution for 100+ concurrent projects, which accelerated environment configuration times by 50%.

SQL analytics, Power BI, and model QA/UAT.

• Developed a complex SQL risk stratification model utilizing taxpayer financial data to segment income behaviours (non-lodgers/omitted income), assigning targeted risk tiers to drive strategic compliance interventions and uncover key trends.
• Architected interactive Power BI financial dashboards for 3 ATO business units, eliminating 10 hours of manual reporting weekly, while upskilling 2 junior data analysts to boost the team's overall reporting capacity by 25%.
• Spearheaded QA and User Acceptance Testing (UAT) for analytical financial models, collaborating with Data Science teams to execute rigorous test cases that identified critical anomalies and improved pre-deployment accuracy by an average of 15%.
• Optimized continuous financial data pipelines utilizing GitLab and Agile methodologies, bridging the gap between technical execution and diverse stakeholder requirements to deliver secure, scalable data solutions.

Database apps and change documentation.

• Analyzed, researched and troubleshooted issues in database applications.
• CRUD operations and data manipulation skills with SQL. Design and implement solutions to avoid recurring problems.
• Used of Azure DevOps with GIT and Jenkins.
• Maintained evidence and documentation for all the changes proposed to recurring problems.