Jan 2024 – Nov 2024 · 10 mos

Jan 2023 – Dec 2023 · 11 mos

• >Added caching concepts to optimize On Demand Hyper tuning methodology enhancing time and cost efficiency handling 100000+ requests a week.
• >Filed two patents for methods around improvement of forecasting accuracy
• >Optimization of code runtime and CPU utilization by 15% and 40% respectively, effectively improving implementation efficiency and reducing operation costs by 10%.

Jul 2021 – Dec 2022 · 1 yr 5 mos

• >Collaborated with solution architect to implement On Demand compute scaling architecture for a robust and cost-efficient forecasting pipeline.
• >Implemented a new modeling method for accounts with inconsistent cash flow, resulting in 5-10% overall higher forecasting accuracy.
• >Consolidated heuristic approach to forecasting for sparse time series data and revamped creation of auto ML processes to improve modeling turnaround time.
• >Developed special pro rata methods for better cash distribution.
• >Directed the initiative of creating smoke and regression tests to maintain code integrity and enhance monitoring as part of the git pipeline.

Sep 2020 – Jun 2021 · 9 mos

• >Strengthened the framework, scalability, and accuracy of the long-term forecasting module.
• >Worked on custom tuning metrics, resulting in a 62% reduction in tuning time at the week level.
• >Defined a ranking system for tuned models to reduce manual effort and facilitate building an Auto ML pipeline.
• >Presented results to the team's global head and wrote an executive summary detailing the value proposition and strategy for presentation to end clients and senior leadership.

Jul 2019 – Aug 2020 · 1 yr 1 mo

• >Formulated heuristic models and automated processes for consistent forecast results using Excel and Python.
• >Incorporated Prophet for long-term consistent and faster forecasting to improve the UAT cycle.
• >Proposed the underlying framework and directory for Treasury cash forecasting and established code conventions for the project.

Jun 2019 – Jun 2019 · 0 mo

• Created and deployed an AI enabled Fin-tech B2B Cloud Application which had the primary role of handling ERP data of the accounts on the UI and provide on demand prediction of payment date for a given open invoice line item.