Data Science and Machine Learning Developer in Cambridge, United Kingdom

Created and deployed a model for predicting when customers would churn or enter in payment default to prioritize the order of the calls. Customers with a higher probability of churn or entering payment default were ranked higher and contacted first than the rest.

Created and deployed a model to predict equipment defects to prevent incorrect and costly replacements. For each maintenance operation at a customer's home, the model recommended the replacement (or not) of the equipment.

Created and deployed a model for recommending the best moment for each customer's next order. The model would output, after each purchase and taking into consideration the customer's profile, the predicted time for the next purchase, which allowed for retention actions or direct marketing.

Created models for sentiment analysis and topic classification leveraging natural language processing techniques and a large number of customer surveys available. The models built allowed to predict the polarity such as love and hate, topics and categories like price, quality of experience, and entertainment, of sentences.

Created a model to predict customer satisfaction based on NPS and VoC scorings. Large volumes of data, corresponding to several customer dimensions, were incorporated into the model in order to have the most accurate predictions possible.

Created a model to predict recurrence of problems after installations or maintenance procedure, allowing to reduce customer attrition. For each new installation and maintenance operation, the model would signal the ones with a higher probability of recurrence such as customer complaints.