Manikanta Sai Parimi

Fares Data (06/2021 - 01/2023)

• Re-architected the Uber Fare Data Ingestion system, transforming a 4 component/services solution into a simplified logging library. This change eliminated 20 billion API requests per day for data enrichment which is 50% traffic over fares store, significantly improving system performance, developer velocity, and maintainability. Done the full implementation and migration end-to-end single handedly over all countries.
• Re-engineered the Uber Trip Fares data generation process, reducing latency from 7 days to just 1 day, enhancing data stability and enabling more timely insights.

Membership exclusives (01/2023 - present)

• Served as the first senior engineer on the team, working closely with a senior manager based in the US to establish and run the team. Led project transitions, managed onboarding for new hires, and coordinated work assignments until a new manager joined the India team, ensuring smooth team operations and continuity.
• Led the DEF+ membership migration track which is currently on going from last 1.5 years
• Worked multiple features like Rides exclusive offers, Priority delivery and Exclusive landing page.

Part of Flipkart Wholesale discovery team

• Designed and lead a team of 3 engineers building the central data ingestion platform for all of Flipkart's wholesale spanning across 4 orgs and making the data available in the Azure data lake for consumption
• Lead the Gateway team of 3 engineers on KTLOs and new features for rendering various pages on the Flipkart wholesale app
• Lead on the generic ingestion platform for building various real-time pipelines like search index generation, transformation jobs etc.

• Lead the development of backend gateway and Orchestration services leveraging Flipkart Poseidon framework, Ngnix for powering the Flipkart Wholesale app
• Designed and developed Backend APIs for powering Product Summary Page
• Built a generic spark library abstracting out the boilerplate code of connecting with sources and sinks easily through config only for HTTP endpoints, Aerospike data store, HDFS and Kafka improving developer velocity.

Worked as a part of Ads team

• Built Audience Manager which is a self-serve platform for business ops teams that is used for generating a specific cohort of users for targeted advertising on mulitple channels like emails, sms, push notifications and on facebook and also on Flipkart website for personalization of ads shown to the users.
• Designed and developed dynamic query generation and segmentation engine leveraging spark for the platform, it scales on datasources rather than cohorts which spans across couple of petabytes of data
• Providing the same data through aerospike store in user path which scales upto 700K QPS keeping the latency under 10ms. Also serving the data in OLAP store like druid for slice and dice analytics

• Part of Ads team
  Worked on Backend APIs for serving reports of Ads performance
• Built Batch pipelines for generating performance and ROI metrics leveraging Hadoop, Hive
• Worked on Product recommendation job which provides recommendations for sellers which has improved the ROI for sellers on Flipkart by ~11.3%