NLP / Product · 2022–2023
AI Title Generator
University of Essex · Group Project
01
Problem
Content teams needed faster generation of relevant titles from source context.
02
Challenge
Train and deploy an encoder–decoder neural title generator with a usable web interface.
03
Solution
Built a neural title generator and deployed it with Flask for interactive inference.
04
Impact
Delivered a working NLP product demo from research to deployable API/UI, closing the loop from coursework to a usable artifact.
Approach
- 01
Prepared and cleaned a text corpus suitable for training an encoder–decoder sequence model.
- 02
Trained and tuned the neural network for title-generation quality against held-out validation text.
- 03
Wrapped the trained model in a Flask API and a minimal web interface for interactive testing.
- 04
Coordinated with group members on model evaluation and interface scope under a fixed coursework deadline.
Outcomes
- Trained encoder–decoder model generating relevant titles from source text
- Deployed, interactive Flask demo rather than a notebook-only result
- Group project delivered on schedule as part of the MSc programme
- Direct precursor to later applied NLP interest in consulting AI-adoption work