Jan 2022 – Oct 2025 · 3 yrs 9 mos

• Search Index Improvements:
 Improved query performance by 20-30% and reduced cost by 10% overall by upgrading the search-index infrastructure.
• Improved the query formation by ~10%, by introducing new Query Engine which uses custom DSL to communicate with search-indexes
• Improved system performance & reduced failure rate due to custom replica-placement, replication and custom operators.
• Infrastructure upgrades to ARM & Apache Solr9
• Data Pipeline (version 2): Next iteration of data pipeline built to solve the scaling problem faced by the existing pipeline.
• Built on Apache Flink for streaming, Apache Druid & Apache Superset for warehousing & Cockroach DB for reporting.
• Dynamic, on-demand scaling of Flink Jobs using custom operator, improves cost of running the pipeline by 30%
• Tiered data storage for various reporting use-cases
• Multiple Services developed to handle support for user-level segmentation & merchandising.
• Service to classify incoming traffic based on user-data using Redis & CockroachDB
• Service to handle classified traffic and apply merchandising based on user-level data
• Improved features set of Schema/Lifecycle management services to reduce failures due to edge cases & support differential update to schema.
• Catalog Ingestion Pipeline (version 3): Highly scalable, event driven data pipeline to handle massive amounts of catalog data from few thousands of products to few hundred million.
• Designed from ground up to handle high rate of partial updates and huge reconciliation steps of hundreds of GB data
• Built entirely on Argo Workflows leveraging Kubernetes’s dynamic scaling along with Spot Instances from AWS, reducing cost of operation by 40-60% depending on time of the day.
• Document Store: Catalog data storage for high speed retrieval when serving the request from index.
• Improved the serving latency by 20% by reducing data overhead within the stack.
• Built on top of PostgreSQL & Aerospike.

Jan 2019 – Present · 7 yrs 2 mos

• Inter Region, Inter Datacenter Network Topology: Design and development of inter-region communication for the services.
• Event Driven Design, using NATS for inter-region communication of configurations, metadata & events.
• Network topology & routing based on index-affinity
• AI Data Pipeline: Designed and developed the data pipeline used for training AI & machine learning models.
• Custom data-lake built on top of S3 & Parquet Files
• Connectors built for easy access from Kubeflow
• Standardisation of Internal Libraries:
• Go (OSS) https://github.com/unbxd/go-base
• Java (Non-OSS) Starter Project & internal functions built on Quarkus

Jan 2014 – Jan 2019 · 5 yrs

• Edge Server:Custom API gateway designed to handle all network related operations, observability and logging.
• Centralised Configuration Management:Hierarchal rule system based on tiers and groups, built on DGraph & NATS
• Query Pipeline:Migration from a single Monolith to Microservices Based Architecture
• Custom Index Solution built on Solr(6.6)’s Fork
• Custom Solr-Plugin, Query Parser, Similarity Class, Search Component etc.
• Other Projects: Some of the notable projects done as different roles.
• Git based CD Pipeline using GitOPS & Flux CD reducing deployment times.
• Internal API Gateway handling inter-team API calls reducing error rates and improving observability.
• Protobuf & GRPC adoption for inter-service communication improving overall latency of internal requests by 20%.
• Multiple service contributions across the stack (Orchestrators, Query Analysis, Promotions, Tracker, Analytics Pipeline v1)

May 2013 – Dec 2013 · 7 mos · Bangalore

• Enhanced front end UI performance by re-modelling in-house PHP framework
• Developed HTML5 mobile app for Android and iOS using AngularJS
• Built CRM tool and Analytics tool for categorizing incoming traffic

Jul 2012 – May 2013 · 10 mos · Bangalore

• Warehouse Management:
• Developed & Maintained multiple versions of SCM product
• Performance tuning. eg: unify multi-stage wave operation to single-stage
• Managed Integration & Build Verification Testing

Aug 2010 – Jul 2012 · 1 yr 11 mos · Bangalore

• (at) London Clearning House:
• Manatained customizations for the client on TCS Bancs
• Enhancements in Equity Management (Custom C++/VMS component)

Jan 2010 – Jun 2010 · 5 mos · Greater Kolkata Area

• Converted an already built ERP on ASP (Application Service Provider) to SAAS model.

May 2009 – Jun 2009 · 1 mo

• Worked on development of Web-Application for NGO for TCS’s Internal Program called Maitree.