Gspann Technologies Pvt Ltd
|Technical Lead
Summary
Led the design and deployment of production-grade Agentic Workflows, RAG pipelines, and multi-turn conversational AI systems for global e-commerce platforms, significantly enhancing search capabilities and reducing operational costs.
Highlights
Reduced LLM API costs by ~35% and maintained sub-2s response latency by designing efficient token-budgeting and response-caching strategies.
Developed a production RAG pipeline, ingesting products into Pinecone with HuggingFace embeddings, achieving a ~60% reduction in search latency compared to keyword-based lookup.
Automated product enrichment via an AI pipeline using LangChain + OpenAI API, reducing manual operational effort by ~40% through auto-generated descriptions and SEO tags.
Enabled natural language queries across 50K+ SKUs by implementing semantic product search with HuggingFace embeddings and Pinecone, replacing traditional keyword matching.
Guided architectural direction, establishing a reusable AI component strategy (prompt templates, shared retrieval layer) and mentoring 6 engineers on modern AI/ML patterns.