Sushma Sharma

AI Solution Architect

General Electric • Vollzeit

Jan 2020 - May 2026 • 6 yrs 4 mos

Revenue & Cost Impact -Identified $0.8M in margin leakage across customer proposals using a Multi-RAG framework -Uncovered $10M+ in cost savings through AI-enhanced "Should Cost" estimation -Delivered a 60% productivity boost for the sourcing team Operational Efficiency -Reduced ETL enhancement efforts by 80% through a common data load framework -Cut ~80 hours/week of manual effort via process automation for functional teams -Reduced maintenance overhead on ~25% of tables by standardising and consolidating redundant data loads -Integrated 25+ source systems into unified data pipelines AI & Agentic Systems Built -Designed a ReAct-based multi-agent LLM workflow that fully automated proposal generation — document retrieval, reasoning, and response synthesis end to end -Architected an enterprise Text-to-SQL platform enabling natural language queries across 200+ structured tables -Built a Multi-RAG framework for extracting and comparing technical requirements across complex energy contracts -Improved retrieval accuracy by 20% using semantic embedding-based query enhancement -Built GenAI risk assessment workflows that automatically surface control gaps from construction site safety documentation -Leading the design of an enterprise knowledge architecture for multi-hop reasoning across distributed datasets LLMOps & Platform -Established company-wide LLMOps best practices — evaluation pipelines, prompt testing frameworks, and model performance monitoring -Designed ML-driven entity resolution workflows to normalise customer records across master data systems Data Science & ML -Built short and long-term financial forecasting models using statistical modeling and trend analysis -Applied clustering models to streamline bulk sourcing by grouping parts with similar specifications -Detected anomalies in T&E data using exploratory analysis and pattern detection, uncovering irregular spending