Senior Software Engineer
CurrentAds Auction team- Rewrite Auto-bid Service: Led a 3-month rewrite project in a 3-member pod, rewriting time-sensitive logic in Golang for improved performance and migrating complex API logic to an existing Django project with relative logic, reducing 95th percentile latency from 400ms to under 100ms and improving uptime by 6%- Search & Discovery ads merge: Co-ordinated and Implemented a 6-month merge plan for technical and data stacks, achieving better quality of code, simplicity, and compatibility, ensuring a zero-downtime transition- New Placements: Developed support for running auctions in pre-search and upgrades page placements, increasing the impression capacity of the auction system by 45M/day- Banner Ads: Implemented support for banner ads in search placement, with an impression capacity of 18M daily impressions, sourced by the marketing team through a Django admin interface- Expansion & Sharding: Integrated two new Demand-Side Platforms (DSPs) and sharded the post-auction data store, enabling microservices to scale efficiently to an increased 6K bids placed per second- Legacy Code Optimization: Removed over 21,000 lines of outdated and obsolete logic from a legacy service over a year, significantly improving maintainability and reducing technical debts and cost of development- Performance Optimization: Profiled, refactored, and tuned a Python microservice designed to predict the probability of app installations in specific contexts (e.g., search results, app details), reducing 95th percentile response time from 200ms to 80ms- Cost Reduction and Simplification: Replaced an Apache Flink cluster with a simple consumer implementation tailored to project requirements to reduce tech stack complexity and maintenance costs, reducing resource usage by 40 cpu cores and 80GB of memory- MongoDB Cluster Maintenance: Scripted toil maintenance of a clustered mongoDB, lowering risks of mistakes and deployment costs
