May 2022 – Jun 2024 · 2 yrs 1 mo · London, England, United Kingdom

• Owned the end-to-end delivery of demand-forecast model to cut surplus fuel inventory, guiding a small team of data scientists and data engineer. It ended up saving GBP 500k annually.
• Led Total Cost of Ownership (TCO) initiative, ran clustering on fleet-transactions to segment drivers and applied ML to uncover 3x savings for fleet managers.
• Developed anomaly detection for fuel-tank sensor data which surfaced accurate levels and made restocking precise.
• Built a tool to highlights key features using drop-column, permutation, and mRMR metrics for any dataset-model pair, improving interpretability and model performance.

Apr 2021 – Mar 2022 · 11 mos · London, England, United Kingdom

• Created the privacy-metrics module to evaluate synthetic datasets, which helped to secure a fintech pilot customer.
• Contributed to the development of the artificial data synthesis module using VAEs.

Oct 2019 – Jun 2020 · 8 mos · San Francisco Bay Area

• Worked in the Brain Networks Laboratory, Department of Radiology & Biomedical Imaging, UCSF, on a computer-vision project that used 3-D structural MRI brain scans to study tissue composition and neuro­degenerative disease.
• Built a U-Net based multi-task learning model which performed 3D semantic image segmentation of brain sMRIs into White Matter, Grey Matter, and Cerebrospinal fluid and the classification task of labeling the brain sMRIs into the following categories - suffering from Alzheimer’s, suffering from MCI (Mild Cognitive Impairment), normal. Achieved a 92% dice coefficient in the segmentation task and 62% accuracy in the classification task.

Sep 2015 – May 2019 · 3 yrs 8 mos · Bengaluru

• Worked on the Cosmos Store Manager team, Azure’s unified storage and analytics stack that manages exabytes of data.
• Revised the node-selection logic for data placement, reducing data-unavailability incidents by 70–80 percent.
• Built an auto-tuning feature that adjusts configuration parameters to available storage and compute, saving engineers weekly manual effort.

Aug 2014 – Jun 2015 · 10 mos · Bengaluru Area, India

• Built services to integrate external payment processors to Amazon's payment platform.
• Contributed to the development of the workflow framework that disbursed payments to the sellers.

May 2012 – Jul 2014 · 2 yrs 2 mos · Hyderabad Area, India

• Part of Oracle Fusion Projects Costing team which was responsible for the development of financial facet of Oracle Projects Portfolio Management Software.
• Contributed to the frontend and the backend development of the account management functionality.

Jul 2011 – May 2012 · 10 mos · Gurgaon, India

• • Customized an ETRM (Energy Trading and Risk Management) software SOLARC according to the requirements of our client Chevron.