Vaibhavi Madhav Deshpande

About

Built and productionized LLM-powered workflows that generate contextual insights from structured financial data, including prompt/retrieval design, data standardization, and reliability controls like rate limiting and batching. Also diagnosed and fixed real-time failures in an automated order validation system using logs/metrics, staging reproduction, edge-case handling, retries, and alerting, while supporting sales/customer teams with demos, scripts, and FAQs to drive adoption.

Experience

Education

Key Strengths

• Took LLM prototype on financial data toward production with end-to-end pipeline design
• Strong data quality and validation approach (multilayer validation, anomaly/hallucination flagging)
• Production reliability engineering (rate limiting, batching, retries, alerting, monitoring)
• Systematic real-time troubleshooting using logs/metrics and staging reproduction
• Effective technical demos/workshops tailored to developer audiences with live code and debugging guidance
• Cross-functional enablement for adoption (demo scripts, FAQs, clear documentation, business-impact translation)
• Deployed LLM system for extracting/validating orders from messy unstructured emails
• Hallucination mitigation via hybrid validation (LLM + fuzzy matching + human review)
• Built RAG pipeline orchestrating multiple LLM agents with LangGraph and MCP
• Reliability-focused agent design with deterministic state transitions and guardrails
• Strong testing/evaluation practice using synthetic + real-world cases and metrics (accuracy, F1, latency)
• Effective collaboration with non-technical stakeholders using prototypes and iterative feedback (hospital admin OCR automation)