Jul 2017 – Jun 2019 · 1 yr 11 mos

• Popularity based ranking for product listings
• Responsible for development and ownership of backend pipeline for computation of popularity of product based on its historic performance.
• Design and development of ranking service which allows more flexibility and experimentation of ranking
• signals, also extensibility to include new signals.
• Design and development of faster ranking refresh of products based on its performance which is computed hourly, which makes ranking of products more dynamic and react faster leading to increase in overall conversion.
• Heavily using spark and cascading to read data from Hadoop/Kafka and making them available to downstream users/data scientists who can consume transformed data from hadoop, reports or queue.

Jul 2015 – Jun 2017 · 1 yr 11 mos

• Demand Ingestion system - Ads platform
• Complete ownership for Ingestion pipleine for PLA(Product Lisiting Ads), streamlining the
• pipeline to be fault tolerant and adding features to increase the fill rate for ads.
• Design and implemention of search for demand entities like images, content etc improving the
• process to create/alter these entities.
• Realtime ingestion of demand entities using storm as processing layer, Elastic search and Hazelcast
• as datastore
• Audience Platform
• Creation of Audience Segments using pre-computed insights like gender, age etc, which allowed
• advertisers, merchandisers to target users for their campaigns.
• Providing reach and supply over audience segments, giving advertisers and merchandisers API’s
• to know the reach(number of users)/supply(views) of their campaign based on audience cuts
• selected.
• Design and development of console, backend API’s and pipelines to provide targeting based
• on segment cuts for advertisers on 3P platforms like Facebook, Youtube etc.
• Platform/Infrastructure Team
• Data Recorder Library - A library to record events and entities with transactionality. Responsible
• for development of relayer which reads data from mysql as events and publishes them to Kafka.
• Nadi - A Framework to transport files between multiple data centres. Mainly responsible for
• tooling, alerting and fault tolerance.
• Job Framework - A framework for processing workflows with batch and stream mode processing.
• Development of API and console which allows user to monitor and control their workflows.
• Technologies used - storm, kafka, mysql, dropwizard, redis, rabbitmq, NodeJs.

Jan 2015 – Jun 2015 · 5 mos

• Develop a library to publish API documentation using swagger generated API doc and publishing it in a tabular form to confluence via a confluence-client.
• https://github.com/flipkart-incubator/Swagger-publish