I build production ML systems — forecasting, knowledge graphs, and edge vision.

Hi 👋! I'm Sarthak Kapaliya.

I am an AI and Data Science professional focused on ML, NLP, computer vision, forecasting, and production-grade MLOps. I enjoy building systems that are explainable, reliable, and measurable. My work bridges the gap between research and real-world deployment.

Recently, I’ve shipped production workflows in insurance analytics (variance analysis, forecasting, knowledge graphs), edge AI for real-time vision, and GenAI-driven document intelligence for complex financial data.

• Mortality experience analytics & forecasting (ARIMA/Prophet) with MLflow and Power BI storytelling.
• Real-time people counting on 200+ edge devices; Jetson Nano + DeepStream; DVC + MLflow for CV models.
• Knowledge graph (Neo4j/Cypher) for GenAI retrieval, boosting chatbot accuracy by 35%.
• LLM-enabled document intelligence; confidence routing + human-in-the-loop for noisy tax PDFs.
• Multimodal AI mock-interview platform with speech-to-text + FER; validated with 40+ users.