Devam Shah

I've built AI-powered data products including DataPilot, a multi-tool LLM data analyst that routes questions across SQL, Python, visualization, chat, clarification, and refusal paths. I deployed it as a multi-user system with Supabase Auth, persistent cloud storage, and Google Cloud Run, achieving 92.9% tool-routing accuracy and 85% execution accuracy.

At Rubixe AI, I applied K-Means and DBSCAN clustering to customer behavior data, supporting targeted personalization and retention strategies. I also built a segmentation pipeline for product recommendations and implemented a data-privacy risk assessment module for PII compliance.

I've also developed Android applications at 3HD Media using Kotlin, XML, Retrofit, REST APIs, Node.js, Express.js, and MongoDB Atlas. My projects include a sentiment-aware anomaly detection platform for 50 Nifty stocks and a CNN-based skin-disorder classification system.