Business Problem: 

A travel recommendation engine wants to provide Travel Personalization and Review Intelligence solutions that enable hotels, hospitality consultants and OTAs to measure and enhance their products & services, improve their reputation and promote guest satisfaction. 

Requirements:​

• Aggregate and analyze online customer reviews to recommend meaningful and actionable insights for the hospitality industry.
• Constantly rank reviews from different sources with varying weightage.

Challenges: 

• Natural Language processing in different languages.
• Building hospitality taxonomy and constantly train models.

Solution: 

• Designed, built and operated core Machine Learning and Lexical Analytics platform using Stanford NLP and custom algorithms.
• Build region specific overrides and multi-tier taxonomy for better accuracy​.
• The underlying ML-Ops platform provides all the necessary tools for data scientists to put their best model into production. 
• Implemented ML models and Deep learning models with experimentation.

Business Outcomes:

• 13 %: Increase in Hotel Occupancy.
• 30 %: More efficient improved revenue per room.

Tech Stack: 

• Custom Algorithms 
• Stanford NLP
• SciKit 
• PyTorch 
• Keras 
• TensorFlow
• OpenCV
• Python
• Docker
• Data Visualization custom charts
• Casandra
• AWS Native

​