Gabriele Gobbi

About

LLM/agentic AI builder who led end-to-end integration of an LLM system into a business intelligence product, creating a scalable, metadata-driven RAG/agent pipeline with an orchestrator that routes queries to specialized agents (including DB-backed quantitative querying). Also built an LLM-to-SQL chatbot and partnered with non-technical stakeholders to capture domain context and improve SQL generation, using automated LLM-based testing to evaluate reliability.

Experience

Education

Key Strengths

• Led end-to-end production deployment of an LLM feature inside a BI product
• Designed scalable, reusable LLM pipelines adaptable across projects via metadata/configuration
• Built agent orchestrator to route queries and select specialized agents/tools (including DB-connected quantitative agent)
• Developed dynamic knowledge bases and RAG workflows to support multiple domains/use cases
• Proactive performance planning during use-case design to avoid bottlenecks
• Automated agent/workflow evaluation using LLM-generated in-context and out-of-context test questions
• Cost-conscious, iterative model/prompt/retrieval selection starting from cheaper models and refining via error analysis
• Effective collaboration with non-technical stakeholders to capture context and requirements (LLM-to-SQL project)