May 2025 – Apr 2026 · 11 mos · Bengaluru

• Working on Ray Core -- https://docs.ray.io/en/latest/ray-core/walkthrough.html

Nov 2021 – Present · 4 yrs 6 mos

Jun 2021 – May 2025 · 3 yrs 11 mos · Bengaluru, Karnataka, India

• I am an Apache PMC and Committer, primarily contributing to Apache Hudi's core transactional engine. Some of the recent key projects include:
• 1. Hudi Connector for Trino - Designed and implemented a native Hudi connector for Trino.
• 2. Asynchronous Indexing in Hudi - A concurrent indexing mechanism without blocking ingestion or other table services running in the background.
• 3. Performance engineering for queries on Hudi tables from different engines such as Presto, Trino and Spark.
• 4. Incremental JDBC Puller - A scalable way to ingest data incrementally through JDBC and handle reconciliation.

Mar 2020 – Jun 2021 · 1 yr 3 mos

• Worked on control plane of Amazon Aurora storage.
• 1. Introduced tiered storage (hot and cold tier) contributing to capex goals.
• 2. Improved failure detection and recovery mechanism ensuring high availability of the distributed storage fleet.

Sep 2018 – Mar 2020 · 1 yr 6 mos · Bengaluru Area, India

• Building mobile payments and financial services platform at scale.
• Tech Stack: Golang, MySQL, Redis
• 1. Distributed Unique ID generator: Developed a fast, thread-safe and highly available UID generator that is unique across time and data centre. The service is meant to be used internally by several other microservices.
• 2. Offline/Online gateway integration: Designed and developed a scalable backend for integration of GrabPay with offline (scan and pay) and online gateways (e.g. iPay88). This has helped in accelerating our merchant acquisition strategy.
• 3. Developed a scalable backend for bulk on-boarding of merchants onto Grab’s payments platform. This has cut down the Ops effort and significantly reduce on-boarding time.

Feb 2016 – Aug 2018 · 2 yrs 6 mos · Bengaluru Area, India

• 1. Optimized the data pipeline saving 43% time and 16% cost (in terms of node-hours of EMR cluster) over previous runs through parallelization of independent and idempotent steps.
• Technology/ Framework: Hive on EMR, Python UDFs
• 2. Designed and implemented the Custom Segment feature, which allows aiq.io clients a one-stop solution to discover audience and create custom segments and activate the same.
• Technology/ Framework: Java, AWS Athena SDKs
• 3. Started contributing to Presto. We are using Athena on top of S3 data lake. Athena's underlying execution engine is Presto.
• Technology/ Framework: Java
• 4. Built a discovery platform from scratch to enable clients to have deeper insights into target users group and their behaviour and thus help them plan better campaigns in the digital ads space.
• Technology/ Framework: React.js, Java, Apache Camel, Vertx.io, AWS Redshift, Memcached
• 5. Enhanced a LexVec word-embedding model based Similar Advertisers API, which takes a brand name as input and returns the most similar brands.
• Technology/Framework: Python, Flask, MongoDB
• 6. Microservices: Designed and implemented segment and inventory metadata search services give a structured information about audience segments and publisher inventories such as their category and users distribution in different geographies.
• Technology/Framework: Spring Boot, Elasticsearch, Hadoop Hive on AWS Elastic MapReduce

Jun 2012 – Sep 2014 · 2 yrs 3 mos · Bengaluru Area, India

• Software development and critical bug fixes for Oracle GoldenGate (OGG), a real-time data integration and heterogeneous database replication technology, and OGG Management Pack, a real-time GoldenGate monitoring system.

May 2011 – Jul 2011 · 2 mos · Bengaluru Area, India

• Developed the web historical reporting client, a very useful tool to streamline call summary at contact centers, using Java technologies with Spring+Hibernate framework.

May 2010 – Jul 2010 · 2 mos · Chennai Area, India

• Worked on algorithm to recover social networks from contagion information.
• Graph Theory, C++