Mar 2023 – Oct 2024 · 1 yr 7 mos · Bengaluru, Karnataka, India · On-site

• Fund Accounting & OTC Derivatives Valuations -
• 1. Transitioned messaging queue from QPid to IBM MQ for better security and performance, involving setup of MQ topics/queues and message publishing adaptation. Also crafted a business-hours-friendly rollback strategy.
• 2. Led the data migration strategy transitioning from Oracle DBaaS to GOS, addressing end-of-support and security updates. Optimised LOB data export/import, cutting 2TB transfer time from 17 to 6 hours, enabling off-hours migration.
• 3. Developed an end-to-end web service to secure database access, involving REST API creation, IDA client/resource registration, and authentication implementation. Configured probe-less round-robin Load Balancer and integrated Tomcat TLS certificates.
• 4. Led Controller Layer upgrade from Spring 3 to 5, transitioning to Annotation-based configurations, enhancing management and development speed.

Nov 2020 – Mar 2023 · 2 yrs 4 mos · Bengaluru, Karnataka, India · Remote

• Apple - Retail Product Catalog
• 1. Designed & implemented a Netflix Graph data model for Apple products, enhancing data retrieval and reducing graph size by 90%, thereby improving server RAM efficiency.
• 2. Created an asynchronous Producer-Consumer pipeline using Redis Stream for efficient processing of data feeds, including new products and promotions, ensuring reliable message retention for successful data handling.
• 3. Performed in-depth load testing, scaling from 1000 to 3000 RPS and employing P95 and average response metrics, which resulted in optimizing the Netflix Graph Schema to reduce API response from 370ms to 7ms and strengthening the unit testing framework to achieve over 80% branch coverage.

Sep 2019 – Sep 2019 · 0 mo · New Delhi, Delhi, India · Hybrid

• Developed an intrusion detection system prototype to identify intrusion (particularly using camouflage) through no-mans land region in India using Renita Net.

Jun 2019 – Jul 2019 · 1 mo · Ahmedabad Area, India · On-site

• 1. Reduced model size from 240 Mb to 4.5 Mb with MorphNet for efficient predictive modeling, and optimized UNet-based Urban Area Segmentation, achieving a mIoU of 0.743 with BOHB.
• 2. Executed LSTM-based NDVI prediction for vegetation cover, using TPE, HyperBand, and BOHB optimization, achieving a MAE of 8.544.

Jun 2018 – Aug 2018 · 2 mos · Coimbatore Area, India

• 1) Developed proof of concept on the Behavioral classification using Unigram/Bigram algorithm
• 2) Implemented ensemble techniques for better performance

May 2018 – Sep 2018 · 4 mos · New Delhi Area, India

• Provided support to teaching faculty by : Clearing the Concepts of Data Structures and Algorithms and Problem Solving