Selected projects spanning churn prediction, demand forecasting, pension risk, NLP workloads, retrieval-augmented generation, and applied AI systems. This page is designed as a portfolio index that connects case studies, demos, and planned systems under one consistent engineering narrative.
A curated selection of projects designed to show breadth across data science, ML engineering, NLP, and AI system design.
End-to-end pipeline to identify at-risk customers and improve retention. Presented as a full ML case study with business framing, preprocessing, evaluation, thresholding, and live prediction flow.
Forecasting engine to optimize stock levels and reduce lost sales. Built as a forecasting case study with lag features, rolling statistics, operational metrics, and live next-day demand prediction.
Collection of NLP components for classification, NER, summarization, and question answering. Positioned as an interactive demonstration of reusable language-processing capabilities.
Retrieval-augmented assistant for domain-specific document Q&A. Designed to combine ingestion, retrieval, grounding, and response generation for practical document intelligence workflows.
Applied AI system for model-driven estimation of building metrics and construction intelligence workflows. Positioned as a product-facing solution rather than a narrow modeling demo.
Risk modeling project for pension portfolios and long-term obligations. Focused on portfolio analytics, forecasting, and quantitative decision support in financial contexts.
Across the full portfolio, the strongest signal is not just model variety, but system thinking: business framing, reusable workflows, architecture awareness, and product-oriented delivery across ML and AI use cases.