Rajani
Maski

• Architected Generative Discovery infrastructure powering RAG-style GenAI applications across 500M+ stock assets from 2M+ contributors, combining dense vector retrieval, hybrid search, and metadata grounding at 2,000+ RPS across hundreds of nodes, with retrieval surfaces in 25 languages.
• Drove infrastructure cost reduction of ~$1M annual alongside 25% latency improvement, while simultaneously improving relevance metrics, decoupling scale from quality tradeoffs.
• Designed and shipped a production agentic query understanding system: LLM-driven intent extraction, query rewriting and expansion, and multi-step retrieval planning integrated into the discovery stack at marketplace scale.
• Built multimodal embedding pipelines supporting image + text semantic search, with MLflow and SageMaker for feature generation and model lifecycle management.
• Designed offline and online evaluation frameworks for retrieval quality using precision/recall, relevance judgments, and human-in-the-loop feedback loops, supporting strict production SLOs.
• Led the Solr → OpenSearch migration of all 500M assets, redesigning the retrieval layer for cloud-native scale and unlocking ML-powered ranking pipelines including Learning-to-Rank.
• Performance work spanning FAISS, vLLM, vector index tuning, and JVM / Lucene profiling, closing the loop between retrieval architecture and the underlying systems.
• Mentor and tech-lead to L2 / L3 engineers on the team for three consecutive years, reviewing designs, driving technical direction on retrieval and ranking workstreams, and growing the bench.

• Designed and deployed vector similarity search systems using Redis; advised enterprise customers on semantic search and retrieval architectures for ML workloads.
• Tuned latency-sensitive distributed systems for production environments with strict SLA requirements at scale.
• Led developer and admin workshops on Redis data structures, RediSearch, and graph workloads.

• Architected enterprise search platforms using Lucidworks Fusion, Solr, and Spark; implemented Learning-to-Rank and ML-based relevance pipelines.
• Designed full-stack search applications for digital workplace and commerce verticals.
• Implemented large-scale search analytics using Apache Spark and Solr.

Search Consultant / Lucidworks India · Senior Software Engineer / Target.com India · Sr. Java Programmer / Happiest Minds · Search Consultant / Sony India · Software Engineer / PointCross Life Sciences · Search Intern / Glassdoor USA