Jon H.

Developed NLP solutions to understand consumer feedback and identify trends from product reviews, social media, and consumer complaints data. These solutions used GPT-3 and open-source Transformer models, such as Bloom & BERT using Hugging Face and PyTorch. Managed a large data science and engineering team at a global consumer goods company. Set up two new data analytics teams in India, Poland, and Mexico with over 40 employees. The role involved recruitment, training, and onboarding of these teams. Worked with several clients to develop their data strategies and product roadmaps and carry out data science maturity assessments. Led a team to develop a machine learning solution for a large government department to prioritize and streamline complex visa applications, saving the department time, effort and money by reducing the strain on the case-working process. Led the development of a new solution to monitor emerging consumer trends across multiple geographies, utilizing natural language processing and time series modelling techniques. Set up a global data science community at a multinational client with over 30 analysts across ten countries. Delivered a forecasting project for one of the largest fast-food restaurants in the world, forecasting menu item sales across all restaurants in the United States and using univariate and multivariate time series models, including ARIMA and AR-Net.

Led the development of a public relations alerting system using natural language processing and time series analysis techniques to alert the leadership for some of the world's largest consumer brands. Managed team-building, industrialized NLP tools for hundreds of users at a large consumer brand company. Supervised a team developing data science reports and dashboards to respond to market research briefs, using social, search, and eCommerce reviews data. Led team building statistical models in Python and R for a UK retail bank covering pricing optimization, customer churn, customer cross-sell, and financial investigations. Analyzed interest rate swaps data and contracts as part of an investigation into LIBOR fixing. Worked with KTrace, a forensic data analysis methodology used to detect anomalies in data known to indicate potential fraud and misconduct. Analyzed security transactions via the stock exchange daily official list (SEDOL) numbers to identify indirect tax savings. Reviewed SQL data warehousing solutions to identify customers affected by mortgage overpayment, mortgage underpayment, and deceased customers. Worked with the National Health Service regulating body Monitor to assess models on patient service costs.