Development of an ontology learning system focused on the detection of interesting terms through a search platform. This project involves cleaning datasets, manual and automated tagging, training of different models related to customer needs, and providing a backend with some endpoints to consume. Generated NLP model to use as a pre-tagger to speed up the labelling process and feed a sense2vec model to associate entities and vectors to find word similarities. Creation of custom scripts for each step of the process to manage different threads reducing the cost of time thanks to the possibility of model training using GPU in a Docker container.
Development of pipelines and machine learning infrastructure involving neural networks with many features, such as time series, buckets, categoricals and numerics. TFX workflow implemented with a custom framework to train different datasets improving the speed of data-focused neural network creation.
Platform developed to monitoring the signaling system and traffic for the Buenos Aires subway.Develop scripts to automate tasks and generate reports.Upgrade multiple services to new technologies.Driver development to integrate data from different terminal devices (PLC, RTUs)