Apr 2022 – Feb 2024 · 1 yr 10 mos

• Search Platform
• Led the development of High Level Architecture (HLA) and Low-Level Design (LLD) for our in-house multi-tenant Search platform, covering real-time ingestion and retrieval layer.
• Designed and implemented multi-tenant services supporting primary business lines, including Food, Instamart, and Dineout serving ~120k RPM at peak.
• Migrated Instamart and Dineout to the platform, achieving an annual savings of approximately ~200k USD and reducing latencies by ~12%.
• Lead a team of 4 developers, managing task allocation, design and code reviews.
• Relevance improvements
• Identified and implemented various improvements for the Food BL and Instamart resulting in ~10k+ and ~5k additional orders/day.
• Neural Search
• Designed and developed support for search on natural language queries using embeddings/vector search resulting in reduced null searches for natural language queries.
• Miscellaneous
• Led technical enhancements in the existing architecture, resulting in improved operational efficiency across multiple facets. Metrics impacted include search latency, latency for discovery flows, Elasticsearch optimization and DDB costs.
• Standardised monitoring and dashboarding, technical specification document and RCA docs.

Oct 2020 – Apr 2022 · 1 yr 6 mos

• Scout
• Designed and developed multi-tenant search service responsible for candidate set generation.
• Service supports search on various entities with multi-dimensionally configurable retrieval strategies.
• Relevance improvements
• Designed and executed multiple experiments on search relevance for the Food business line. These include semantic search, field normalisation, slot popularity, etc. Critical search metrics impacted include increment in avg order value(~1.8Rs) and orders(~5k/day).
• Devised and implemented v0 approach for classifying results as high/low confidence on the Food dish search page, improving search relevance and additional ~2k orders/day.
• Miscellaneous
• Led technical enhancements in the existing architecture, resulting in improved operational efficiency across multiple facets. Metrics impacted include search latency, search corpus recovery time and Elasticsearch optimizations.
• Identified and fixed issues in the Ingestion pipeline reducing cron duration(~55%) and storage(~60%).

Nov 2019 – Oct 2020 · 11 mos

• Voyager
• Designed orchestrator service responsible for query enrichment, ranking, filtering, deduping, ads enrichment and serving ~100k RPM.
• Facets and filters
• Designed and developed framework for generating facets and filters dynamically depending on the result set.
• This also enabled contextual search which powers category based search.

Jun 2018 – Nov 2019 · 1 yr 5 mos · Gurgaon, India

• Insights Engine
• Designed and developed Insight Engine, a service to convert raw data into insights and actions.
• Used to generate dashboards consisting of widgets which can be dynamically controlled.
• Developed ETL pipelines using Airflow to dump data from Hive to Elasticsearch.
• OpsGateway + Monolith to Microservice Migration
• Part of developing a microservice which acts as a gateway to all our microservices and also contains legacy code which was extracted from a monolith service.
• Ratings and Room Recommendation Service
• Developed a microservice which calculates rating for a property based on various parameters which then decides the search listing.

Jan 2018 – Jun 2018 · 5 mos · Gurgaon, India

May 2017 – Jul 2017 · 2 mos · Bengaluru Area, India

• Part of BHIM-1.4 backend development team. Responsible for creating new API's and improving code performance.
• Technologies used were: NodeJs, PostgreSQL, Redis