Mar 2025 – Sep 2025 · 6 mos

Mar 2022 – Mar 2025 · 3 yrs

• As a part of the Search Semantics team:
• Worked on designing, developing and evangelizing of the retraining platform used for regular retraining of the models present within the ecosystem, responsible for processing of more than 5 TB of data and deploying in critical user paths with zero downtime.
• Worked on optimising the search graph database, and developing strategies to shard the same to scale beyond 30B edges and 60M nodes.
• Worked on designing and developing solutions to increase relevant recall of the system of various BUs like Grocery, Mobiles, and revamp of the UI/ UX as well of the same. This resulted in increase in search CTR by 50 bps, decrease in null searches by 70 bps,and increase in revenue by 24 bps.
• Worked on identifying and solving problems through the use of generative AI to improve the query understanding in critical user paths. This resulted in increase in revenue by 0.3%, search CTR by 0.2% and decrease in search pain by 10 bps.
• Worked on scaling of semantics services to handle upto a QPS of 80K for various BBD editions, Flipkart's flagship annual sale event, and also mentored various folks for the same.
• Enhancing the SDLC processes within the team by writing coding practice guidelines, integration with SonarQube, and enabling tech sharing across the team.
• Collaborating with various functions like Product, UI/ UX, Partner teams (internal and external to Search), Business teams to drive various projects right from requirements gathering to taking them live.

May 2021 – Mar 2022 · 10 mos

• As a part of the Search Semantics team:
• Worked on improving the spell correction module by solving for various types of query error classes ex: alphanumeric queries etc. which resulted in increase of spell accuracy by 2%, decrease in search pain by 20 bps, and % users saying irrelevant results by 0.98%.
• Worked on designing and developing distributed caching for the spell service, which results in more than 20% latency drops during peak traffic hours.
• Acted as a SPOC to scale the semantics services to around 80K QPS for the BBD, Flipkart's flagship yearly sale event.
• Worked on designing (HLD and LLD) of the retraining platform for regular retraining of the various models present within the search ecosystem.

Mar 2021 – May 2021 · 2 mos

• - Ideated personalisation in search ranking.

Aug 2019 – Feb 2021 · 1 yr 6 mos

• As a part of Search Platform Team:
• Worked on building self managed ElasticSearch:
• a. Built service to modify ES (instance type, EBS storage, number of nodes, ES version etc.), without any search or indexing downtime.
• b. Automated various parts of infra as IaaC.
• c. Ran various load tests for self managed ES.
• d. Developed code within search application to support migration to self managed ES from AWS managed ES.
• e. Helped various other teams within Traveloka to migrate to Self Managed ES.
• Developed an highly asynchronous auxiliary pipeline within Indexing Service, which resulted in 50% better latencies and 35% less primary Database CPU.
• Redid the caching layer of Search Service.
• Representing Search Team, by acting as SPOC with different product teams.
• Re-architecting of search as a service.

Jul 2018 – Nov 2018 · 4 mos · New Delhi Area, India