May 2022 – Aug 2022 · 3 mos · Sunnyvale, California, United States

Jul 2018 – Aug 2021 · 3 yrs 1 mo · Bangalore

• Contributed to various components in containerized microservices architecture, added new features which included providing GUI, adding REST API endpoints, generating and parsing service configuration, collecting and storing real time data, and doing analytics. Below are some of notable work items
• Researched various time series databases and implemented an architecture to extend single node TSDB to multi-node cluster based database, thus improving performance many folds by providing scalability and high availability.
• Created a customizable dashlet that displays timeline view of health events generated by telemetry application for better understanding of events behaviour over past.
• Developed a daemon to publish live telemetry data and notify alerts on detection of network issues to various destinations like kafka, rabbitmq, slack, webhook, etc.
• Implemented a collector for Inmon sFlow protocol in golang and integrated it with other services, thus providing detailed and quantitative traffic measurements at gigabit speeds.

Jan 2018 – Jun 2018 · 5 mos · Bangalore

• Worked in Paragon Insights product formerly known as Healthbot, It is a network health monitoring app that collects, monitors real time sensor data and notify failures to users. I wrote many automated rules/configuration to subscribe sensors and take actions when anomalies are detected. I got exposure to time series database Influxdb, open source data collector Fluentd, and to the world of docker containerization & micro service architecture.

Jan 2018 – May 2018 · 4 mos · Bengaluru Area, India

• Researched on Scene understanding i.e generate an image given the text description of the image using GAN (Generative Adversarial Network). Research objective was to show how GAN can be used in the fashion industry to generate attire images given its text description and empower fashion designers to bring out complex and abstract thoughts.
• We scraped Indian fashion attires with description from web and trained a model on it. Trained Model was able to synthesis many plausible visual interpretations of a given attire description.

Jan 2017 – Jun 2017 · 5 mos · Bengaluru Area, India

• Researched KPIs that can be used to effectively measure power consumption of a system and identified important Hardware Performance Counters that can help us achieve this goal. Formulated an equation that gives similar power usage as compared to already existing methods with significantly lower implementation cost. Proposed a methodology to meter power on Openstack that can be used in billing based on resource usage.