Artificial Intelligence Engineer
Current•Query rewriting: Training a model A to identify whether two queries are paraphrases with traditional classification algorithms. Then generating a big training data based on this model A. Finally training a transfomer model with the generated big training data as the final model B. This model B is similar to a translation model, which “translate” the original query into another one or several queries to improve the accuracy, recall, versatility of web searches.•Elo recommendation system: Kagglers develop algorithms to identify and serve the most relevant opportunities to individuals, by uncovering signal in customer loyalty. This is a traditional regression problem which involves a lot of feature engineering and traditional regression algorithms •Homecredit classification task: Homecredit makes use of a variety of alternative data--including telco and transactional information--to predict their clients' repayment abilities. This is a traditional classification problem which involves a lot of feature engineering and traditional classification algorithms such as linear regression, gbdt, etc.