Aaron Lombardi Email and Phone Number

The Core ML group is part of the Applied Machine Learning org in PSS. Our charter is to advance machine learning through applied research in event prediction, personalization, deep learning, natural language understanding and large scale ML.What do we do?- Advanced ML/AI algorithms: Distributed ML, Neural Networks, Deep Learning for Text understanding and others.- ML solutions: Modeling ML problems, driving internal adoption of advanced ML techniques and metrics.- Advanced… Show more The Core ML group is part of the Applied Machine Learning org in PSS. Our charter is to advance machine learning through applied research in event prediction, personalization, deep learning, natural language understanding and large scale ML.What do we do?- Advanced ML/AI algorithms: Distributed ML, Neural Networks, Deep Learning for Text understanding and others.- ML solutions: Modeling ML problems, driving internal adoption of advanced ML techniques and metrics.- Advanced ML systems: Large scale ML training and prediction systems (PSSLearner Flow and AML Predictor)Who are we looking for?- ML Managers: Lead the CoreML algorithms and AI group and develop next generation algorithms for PSS.- ML practitioners and researchers: Research Scientists and hard core ML engineers.- ML systems & product engineers: Systems and product engineers with prior experiences in Big data, data visualization, large scale systems and ML infraExample projects:- DeepText: State of the art text understanding leveraging Deep Neural Networks. Handles more than 300,000 queries per second across 20+ languages.- PSSLearner Flow: PSS Learning system used by hundreds of engineers and ML practitioners to build ML training pipelines. Currently we train tens of thousands of models every week. Used by more than thirty teams in PSP.- Sigrid: Large Scale SGD algorithms and system capable of learning from billions of examples of data and training models with billions of weights. Currently used by Ads, Feed and Search.- Video Compression: This is an example of a comprehensive ML solution driven by the team. The system predicts the best parameters to be used to compress videos while minimizing quality loss. Reduces the network bandwidth used in rendering videos across millions of devices every day. Show less

Current: Speech Recognition - Large vocabulary neural network language modellingMaximum EntropyDeep Learning / Neural NetworksPreviously:Display Advertising Reporting Infrastructure - Tech Lead Report Scheduling InfrastructureLead scheduler from initial design discussions through development to rollout with first product and continuing evangelizing and seeking out more customers.