Extracted content from any URL with Crawl4AI, converting unstructured pages into clean, usable text for downstream AI processing.
Vectorized the extracted content and stored it in FAISS, overcoming open-source model context limits and enabling fast, accurate retrieval across large pages.
Summarized webpages and powered RAG-based search with OpenAI/Ollama + DeepSeek, delivering deeper, context-aware answers rather than simple summaries.
Enabled interactive chatbot Q&A through a Streamlit interface, creating a practical research tool that can cut manual analysis time by up to 35% while improving insight discovery.
Made with Bullet