At ValueLabs, I architected an enterprise conversational analytics platform that turns natural-language queries into reports, dashboards, and business insights through agentic AI workflows. The platform was adopted by five enterprise customers and selected for demonstration at the company's annual technology event.
I formulated a hierarchical summarization algorithm inspired by K-Means clustering to process datasets exceeding 1–2 million tokens with an internally hosted LLM and an 8K-token context window. I also engineered LangGraph and DSPy workflows for planning, tool orchestration, query decomposition, intent routing, and response generation.
I led the architecture and implementation of an AI-powered platform for automating database migrations across different technologies. Its schema redesign pipelines use ClickHouse, Amazon DMS, Anthropic LLMs, FastAPI, and React, and achieved approximately 70% accuracy during enterprise validation.
My other ValueLabs work includes a product classification solution that achieved approximately 75% accuracy, evaluating machine learning models for insurance claim settlement prediction, and developing LLM-powered compliance and FAIR validation solutions. I work with Python, machine learning frameworks, and cloud platforms including AWS, Google Cloud Platform, and Azure.
