QBE European Operations forms a significant part of the QBE Insurance Group, which is one of the world’ s leading insurers and reinsurers, with operations in 36 countries worldwide. In QBE I am responsible for:
• Lead Text Analytics and NLP applications that enhance fraud claims detection
• Communicate with claims department to identify new cost-saving opportunities utilising NLP techniques
• Promote and apply recent deep learning models to enhance existing projects
• Recruit new team members
SwiftKey (now part of Microsoft) provides next generation mobile text entry system based on artificial intelligence and natural language processing technology.
I apply NLP and machine learning techniques to design, implement, evaluate and improve our text prediction accuracy and language model efficiency. Most of my projects are centred around:
• Design and experiment Neural Network language models for complex languages, e.g. Chinese
• Experiment with Character/Morpheme models for complex/highly inflected languages
• Evaluate and analyse performance metrics to understand user behaviours
• Apply supervised classification techniques for automatic language classification
• Apply unsupervised NLP and ML techniques to discover unlisted words
• Continue experiments on data and extract features that can improve our models
I utilise Hadoop MapReduce and Spark distributed computing for our machine learning components, as well as CNTK for deep neural net models.
Fizzback (acquired by NICE systems) is a leading software provider for Real-Time Customer Feedback and Customer Experience Management (CEM) solutions.
In Fizzback I am responsible for researching and developing the next generation NLP modules. They include automatic spelling error detection and context-based correction, topic classification and clustering as well as customer reviews sentiment analysis.
Research engineer in Yahoo! Answers team based in London. Yahoo! Answers is a community-driven Q\&A web service that allows anyone around the world to ask questions and to answer questions by other users. Currently it attracts approximately 20-million unique users and 40-million page-views per day.
My responsibilities include maintaining and enhancing Yahoo! Answers natural language processing/machine learning research components such as:
- Implement and expand the multi-lingual automatic text classifier
- Question content similarity measures
- Design algorithms for delivering relevant content to relevant users
- Hadoop MapReduce for data mining
- Experiment and evaluation for content/user behavioural analysis with HDFS data