Jul 2024 – Jun 2025 · 11 mos · Toronto, Ontario, Canada · On-site

• Team: AI Platform (Semantic Search)
• Delivered key features to power RAG workflows for internal GS documents, significantly enhancing LLM integration and semantic search capabilities for our teams’ customers. Major contributions include:
• Built on-demand replay scripts for a vector database's document ingestion pipeline, enabling re-ingestion of documents, replay of embeddings to vector stores from S3, and copying embeddings between vector databases.
• Developed a reconciliation service to maintain consistency across Blue/Green vector DB clusters by detecting and resolving ingestion or deletion failures.
• Led self-service schema onboarding, reducing manual onboarding time by a few hours per schema, eliminating human error and accelerating team workflows.

Jan 2020 – Jan 2021 · 1 yr · Bengaluru, Karnataka

• Worked in the Funding Liquidity Execution (aka FLX) team, major contributions:
• Automated currency position reporting, eliminating manual intervention.
• Delivered foundational components for a cash projection system, reducing FX hedging costs and volatility across EMEA and USD.

Jan 2021 – Jun 2024 · 3 yrs 5 mos · Toronto, Ontario, Canada

• As one of the founding members of the Ad Serving Team at Nextdoor, I've built several core features for Nextdoor's proprietary Ad Server from scratch. Major contributions:
• Benchmarked and optimized Elasticsearch for ad targeting index; built an in-memory inverted index, saving ~$20K USD/year while maintaining sub-5ms latency across 200K+ ads.
• Reduced Redis storage costs by 97% through Protobuf-based optimization for ML model feature storage.
• Spearheaded multiple core ad platform features: ML model A/B experimentation, unified ad taxonomy, bid multipliers, and frequency capping.
• Optimized DAO layer for Nextdoor Ad Centre, reducing server start-up time to a few minutes from over an hour.
• Accelerated team velocity by mentoring newly onboarded colleagues as well as interns.
• Key Projects Delivered: Ads Taxonomy, Bid Multipliers, Ads Targeting Indexing and Retrieval, Feature Store for ML Models, Frequency Capping, A/B Experimentation Setup for ML Models

Jan 2021 – Jan 2021 · 0 mo · Bengaluru, Karnataka, India

• As a part of the Job Ingestion team (within LinkedIn Talent Solutions) I:
• Developed custom job scrapers for major LTS clients (e.g., Greenhouse).
• Migrated internal tool JAM’s search system from Oracle to Galene (Lucene-based), preventing timeouts for complex queries.

Mar 2019 – Jan 2020 · 10 mos

• Major contributions in the Ad Serving team:
• Designed recommendation strategies for geo-targeted and contextual ad delivery.
• Integrated TensorFlow model pipelines for real-time bid correction.
• Reduced contextual ad-server latency by implementing lazy k-way merge collections.

May 2017 – Feb 2019 · 1 yr 9 mos

• Built core features for Times Internet's first in house Ad Server

May 2016 – Jul 2016 · 2 mos · Noida Area, India

• The internship role included hands-on working within the Ad-tech department (Colombia) at Times Internet. Two projects with which I was associated were: creating JUNIT based integration tests and a Spring-framework based Web Application.