Wallapop is a mobile marketplace whose items are defined by a description and a photo uploaded by the seller. A full-time position is open at Wallapop to develop an automatic system for the classification of items based on their textual description. To classify an item, it must be first represented in a vectorial space. Two types of representations are considered: a representation based on traditional vector space models, created with words or other linguistic features, and neural-network based models like word2vec, doc2vec or convolutional neural-networks. A comparison should be run between various representations to find the best one that allows a classifier to automatically detect the type of items on sale.
The ideal candidate should be a Computational Linguist with experience or interest in short text analysis and deep learning. Strong programming skills are required in Python, knowledge of Theano framework, and an adequate knowledge of the Spanish language.