Real estate professionals and market analysts often need to synthesize information from multiple long‑form news articles and reports to derive actionable insights—a task that is time‑intensive and prone to oversight. To address this challenge, I designed and developed PropWise, an AI‑powered research assistant that leverages cutting‑edge language models and retrieval‑augmented generation to extract and summarize domain‑relevant information from real estate news URLs. This tool streamlines the research process, enabling users to obtain precise, source‑cited answers to custom queries without manually reading multiple sources.
I implemented a RAG (Retrieval‑Augmented Generation) pipeline using LangChain that ingests multiple article URLs, preprocesses and scrapes content, and stores vector embeddings within a Chroma DB vector store. The system retrieves relevant segments based on user‑provided questions and employs a large language model (such as Groq’s Llama‑based model) to synthesize coherent, context‑aware responses, complete with source citations for transparency and verification. I built an intuitive Streamlit interface for users to input URLs and queries, and containerized the application using Docker to facilitate reproducible deployment.
PropWise demonstrably reduces research time by automating the extraction and contextual summarization of insights from multiple real estate news sources. The RAG‑powered workflow ensures answers are both relevant and traceable to original content, improving credibility for analysts and decision makers. This project highlights competencies in NLP, RAG architectures, LLM integration, vector databases, interactive UI development, and containerized deployment—all of which directly align with roles in applied machine learning engineering and AI solutions development and are evidenced in the repository’s codebase.