Making roads safer through Machine Learning
In our B2B business model continuously improving on our clients’ models requires frequent model re/training as large amounts of new sensor data becomes available. To support rapid research and development, flexible, scalable and trackable training pipelines are required.
In this talk we will show how to use Hydra, Optuna and MLfow to create training pipelines in Python which are easily configurable, maximise model performance and can be traced.
Principal Data Scientist II, Machine Learning Guild Leader, Cambridge Mobile Telematics
István Barra is a Principal Data Scientist at Cambridge Mobile Telematics (CMT), where he works in the Crash and Claims team. Before joining CMT he held various data science related positions in finance and mobile gaming. He has a Phd in financial econometrics form the Vrije University Amsterdam. His research was published in scientific journals such as Journal of Applied Econometrics and Journal of Business and Economic Statistics.
Richárd Nagyfi is a Senior Data Scientist at Cambridge Mobile Telematics, working on Crash Detection and Research and Development. Previously he worked on movement based diagnostical ML models, NLP for smart assistants in Hungarian and VR methodology and he has a keen interest in cutting edge research. Is a PhD candidate at ELTE.