Jan 2021 – Jun 2025 · 4 yrs 5 mos · San Francisco Bay Area

• Raised 170M to build the future of insurance.
• Nirvana is modernizing the entire insurance experience - using technology to help customers have safer outcomes while also rewarding them financially for being safe. We’ve combining the best of human expertise and AI to understand each customer uniquely, and rewarding the best ones for being safe - this is a fairer outcome for those companies who rise above the average.
• This means we can provide world class insurance, help customers get safer and help them drive better bottom line results.

Mar 2017 – Jan 2021 · 3 yrs 10 mos · Palo Alto, California, United States

• 2020-2021
• Overall lead for the Data Platform team and founding member of Turbo - the Performance and Observability team. We built and maintained an in-house TSDB backed by Apache Druid, and a petabyte-scale data lake processing engine atop Apache Spark, to power observability at Rubrik. From business applications, distributed services, and down to the Linux kernel, we use bleeding-edge technologies at scale to collect and generate real-time, actionable insights from thousands of clusters generating petabyte scale data each. From business applications, distributed services, and down to the Linux kernel, we used bleeding-edge technologies at scale to collect and generate real-time, actionable insights from thousands of clusters.
• 2018-2019
• Led a 20+ eng team (largest team at Rubrik at the time) to build the industry’s first ransomware mitigation data security product. The product allows detection and near-zero RPO recovery for critical infrastructure.
• Architected the data path to intercept and ingest I/O in real-time with minimal overhead. Designed a bufferless protocol to reliably handle transient network saturation. Worked with the distributed filesystem team to build QoS support for latency-sensitive workloads.
• 15+ granted patents for this product line
• 2017
• Led the development of Rubrik’s first data protection product for private clouds. Re-engineered core parts of Rubrik's software defined data management platform to support the product at scale.
• Built the first version of Rubrik CloudOn, a cloud migration product to move workloads from data centers to clouds.

Sep 2016 – May 2017 · 8 mos · Bangalore

• Led development of "100% Itemization", which increased Bidgely's disaggregation coverage for household electricity consumption from 60% to 100% by blending statistical modeling and sensor-based ML.

Jan 2016 – Jan 2016 · 0 mo · New Delhi

• Bite-sized casual learning app with ML generated content graph

May 2015 – Jul 2015 · 2 mos · Mountain View, California, United States

• Developed and implemented an algorithm to estimate energy consumed by lighting appliances in a household using only the aggregated energy usage data of a residential building.
• An improved version of the algorithm is in production and serves a majority of Bidgely's customers.