Jan 2024 – May 2024 · 4 mos · San Francisco, California, United States · Remote

Aug 2023 – Dec 2023 · 4 mos · Tempe, Arizona, United States · On-site

• Teaching Assistant for CSE 412 (Database Management)

Jan 2023 – Aug 2023 · 7 mos · Tempe, Arizona, United States · On-site

• Unified Compilation Framework for integrating SQL and deep learning inferences using Velox

Feb 2022 – Aug 2022 · 6 mos

• Improved Redis throughput by 3.25x by migrating to Redis 6.0 and switching to Lettuce client
• Implemented pipelining and manual flushing of commands to improve Redis performance
• Migrated dead letter topics from Kafka to Apache Pulsar (as a managed service) further relaxing mainline throttling
• Ongoing:
• Developing a novel approach for retrying failed messages at dynamic intervals(up to 24hrs) in a high throughput environment on Apache Pulsar
• BCP-DR: Making the messaging system active-active across multiple DCs

Feb 2019 – Feb 2022 · 3 yrs

• Highly scalable Message Bus (set of 5 micro-services) - At least once, ordered delivery of messages at high throughput, with fixed interval retries and on-demand consumption of failed messages
• Rearchitected the messaging platform and removed redundant components which lead to tighter guarantees around message ordering, improved latency for the consumption of failed messages by 5x, reduction in resource usage by 33%, and reduced on-call load. This also helped provide a more predictable way for storing failed messages
• Re-architected the messaging platform and implemented dead letter queues that slowly drain failed messages to the sideline store thereby acting as a buffer. This lead to relaxed throttling in mainline consumption by at least 6x and reduced load on the sideline store which was a single point of failure.
• Subscription allocation improvements :Improved the strategy for allocation of subscriptions across consumer apps to ensure zero downtime of messages consumption and better utilisation of resources (20% reduction in resource usage). Reviewed and implemented distributed locking using curator recipe thereby improving the subscription allocation state management.
• Enabled all micro-services to recover gracefully in case of data store failures (MySQL, Kafka, Redis, zookeeper) so that apps do not have to be restarted thereby reducing on-call load
• Migrated apps from VM to Kubernetes
• Designed and implemented workflows to make the messaging platform SOX compliant. Added Authentication and authorisation workflows to restrict access to resources
• Migrated the messaging platform from one data centre to another with 0 downtime of production and consumption workflows. Performed NFRs in the new data centre to ascertain the VM types to support the data store and stateless apps

Apr 2018 – Feb 2019 · 10 mos · Bengaluru Area, India

• Re-architect monolithic Samsung Account Service used for global Samsung user accounts to a set of
• micro-services(serving 10k QPS); Migration of legacy systems to Amazon Web Services
• Designed and implemented a Key-Value as a service for storing IoT device data
• Developed an admin portal with Authorization enabled(OAuth 2.0)
• [Research] Real-Time audio classification into household sounds(dog-barking, baby-crying, etc)
• Captured audio from Samsung surveillance camera’s RTSP stream and isolated impulses from the feed
• into audio clips. Extracted features such as MFCC from the audio clip and fed the features to a Hidden
• Markov model for classification
• Rule-based anomaly detection system for data generated by IoT devices

Jun 2016 – Mar 2018 · 1 yr 9 mos · Bengaluru Area, India

• • Audio using supervised Machine Learning Techniques : Detected audio impulses from rtsp feed, extracting relevant features like MFCC from audio samples and classifying them using SVM technique

Jun 2015 – Jul 2015 · 1 mo · Bengaluru Area, India

• Studied about the various cyber attacks
• Provided working solutions against SSL Strip and Evil Twin attacks