Google Search

As part of Google’s Search India Team, built exam OSRPs and the interactive Exam Quiz, reaching 100M+ educational users on a product serving 1B+ queries per day.

Migrated Google Search’s serving query-understanding architecture from entity-driven to intent-driven through the diOSRP Migration, enabling better Organized Search Result Pages (OSRPs) that answered user intent directly instead of routing users to the traditional 10 blue links; impacted 0.05% of Search traffic.

Launched localized Exam OSRPs in Hindi, Tamil, and Telugu, adding exam dates, fees and waivers, tabbed result layouts, and the Periodic Table. Localization reached 0.03% of Search queries and 0.0019% of educational queries, improving accessibility; also designed a scalable Language API for Indic locales.

Contributed across Google Search’s monorepo, one of the world’s largest codebases with 2B+ lines of code and 10M+ files, using C++, Java, Python, Go, gRPC, Protocol Buffers, Knowledge Graphs, and other internal technologies.

Collaborated across Search teams to resolve production issues, align requirements, and prioritize high-impact improvements.

Contributed to Google Bard’s early model-improvement journey by training and evaluating LLM responses, helping improve output quality as the technology evolved from unreliable early behavior toward practical usefulness.