Text Data Insight Engine

I have a growing collection of unstructured text—support tickets, survey comments, and internal reports—that needs to be transformed into clear, actionable insights. I want an end-to-end AI solution that can ingest this material, clean and preprocess it, then surface patterns I can rely on for decision-making.

Here is what I have in mind: an NLP pipeline in Python that handles language detection, tokenisation, stop-word removal, and vectorisation, followed by modules for sentiment analysis, topic modelling, and keyword extraction. Named-entity recognition would be a bonus if it improves the overall insight quality. The core model can be built with spaCy or Hugging Face Transformers; I am open to whichever framework best balances accuracy and runtime. Results should be delivered as both exportable data (CSV or JSON) and a concise visual summary—Jupyter notebooks, Streamlit, or a lightweight dashboard are all fine.

I will provide a sample of the text corpus at project start. Your deliverable is the working code, a requirements.txt or environment.yml, and short deployment instructions so I can reproduce everything on my side. Clear, well-commented code and a brief README will be part of the acceptance criteria.