Saptarshi Bandopadhyay

• Engineered an AI assistant for Outlook to process hundreds of incoming emails per hour, automating meeting scheduling, email responses, follow-ups, rule-based delegation, and post-meeting note generation. • Designed a 9-stage LangGraph workflow for warehouse compliance, integrated with Microsoft Graph webhooks, to process email conversations between managers and warehouse workers, classify 6 operational workflows, extract structured compliance data, track workflow states, and generate risk-analysis reports. • Engineered an event automation platform with FastAPI, MySQL, and SQLAlchemy, orchestrating 3 workflows for attendees, sponsors, and partners across hundreds of registrations per event, with form-triggered processing, event matching and scheduling, automated email routing, onboarding, follow-ups, and database-backed workflow tracking.

• Architected a multi-step LangGraph RAG agent over a corpus of tens of thousands of documents, using Milvus hybrid search and streamed responses with query expansion, query decomposition, and answer regeneration. • Built an autonomous AI-ETL agent that scraped and transformed 10,000 records/day, automatically processing unstructured sources for PostgreSQL ingestion.

• Extended the NumPy frontend of the Ivy framework by implementing an instance method and developing comprehensive tests, contributing changes through a merged pull request.