Feb 2021 – Oct 2021 · 8 mos

• Trust & Safety Team (Fraud Detection)
• 1. Working on enhancing fraud detection services in the team
• 2. Drove multiple service migrations across the team.
• 3. Working on easing and automating Cassandra maintenance operations

Apr 2019 – Feb 2021 · 1 yr 10 mos

• Trust & Safety Team (Fraud Detection)
• 1. FeatureStore
• Part of the team responsible for building a highly scalable multi-tenant platform called FeatureStore that manages thousands of features for training and online serving of machine learning models and rules engine for fraud detection.
• Designed the catalog service which maintains all meta information for the features and is the central piece of the platform. Worked on cli and flask powered UI for easier client interactions with featurestore, wrote a SDK which any client can integrate to ingest features into featurestore
• Worked on the serving and ingestion layers for the store.
• Wrote spark powered pipeline for batch feature ingestion into the store. Around 80 batch jobs on a daily basis ingest cumulatively billions of features in the store.
• Worked cross team to integrate stream (real-time) feature ingestion into the store
• Worked closely with Aerospike NoSQL database to establish storage contracts and optimizations for the store.
• The system ingests around 30 billion feature values on a daily basis and serves a million features per second during flash sale
• Ownership of the platform. All code changes, new feature asks etc. go through me
• 2. CDC (Change Delta Capture)
• Part of the team responsible that worked on a service called CDC which comprises of generator, commit manager and consumer components
• It integrates with internal teams' to read data from the sources and propagates only the changed data from the last execution effectively bringing down the data volume. Clients can consume this data and define their processing logic on top of this.
• Designed commit manager component end to end
• Worked on the consumer component
• For one of clients which used to ingest 200M records on a daily basis, this service brought down the number of records to 10M

Jun 2017 – Mar 2019 · 1 yr 9 mos · Bengaluru Area, India

• The work involves two major roles:
• 1. Contributing towards the Code Evaluation Engine used by HackerEarth.
• 2. Managing the challenges conducted on HackerEarth which involves problem creation and
• problem testing as the main parts of the process.

Jun 2016 – Dec 2016 · 6 mos · Bangalore

• My responsibilities include:
• 1. Making good quality content for the HackerEarth website. I set and test algorithmic problems for challenges conducted on HackerEarth.
• 2. Writing tutorials on various algorithms and data structures used in competitive programming.