Harshil Gupta

Architecting and deploying production AI/ML data pipelines and LLM-powered workflows, driving intelligent automation across business units. Leading cross-functional data engineering initiatives integrating Generative AI into enterprise data infrastructure. Building scalable Python backends and orchestrating multi-model AI systems in cloud environments.

Engineered a scalable LLM-based multi-label classifier using GPT-4, Gemini, and Mistral, automating sentiment, user-type, and two-level reason tagging from user feedback — reducing manual review time by 85%. Analyzed 12 months of historical user feedback (300,000+ records), optimizing data collection touchpoints and improving coverage by 40%, surfacing recurring platform pain points and UX gaps. Achieved 90%+ classification accuracy across diverse feedback types using a robust Python pipeline with multi-LLM ensemble evaluation and structured prompt engineering. Integrated a CrewAI-based multi-agent framework for parallel sentiment, reason, and user-type analysis, boosting processing efficiency by 60% and compressing feedback loop time from weeks to under 24 hours.

Migrated 50+ legacy reports to Snowflake, writing and optimizing complex SQL queries and applying arithmetic transformations to ensure data accuracy and consistency across business units. Designed and deployed a Streamlit-based Python Data Analytics tool enabling self-serve reporting and interactive data exploration for non-technical stakeholders. Built and maintained Power BI dashboards consumed by 5+ internal Business Units, aligning KPI visualization with stakeholder requirements. Coordinated data requirements with internal BUs and stakeholders across 3 simultaneous projects in a team of 4, ensuring on-time delivery and cross-functional alignment.