Spyros joined Amazon ahead of the initial Echo launch, working to improve Alexa’s far-field speech recognition accuracy through careful data selection and processing, and speaker adaptation. He then shifted focus to natural language understanding, and is now developing a spoken language understanding architecture that enables developers around the world to build high-quality user experiences in Alexa.

Why join the Alexa Machine Learning team? The reasons are many, Spyros says. First, the team has the opportunity to make a positive impact to the lives of a large set of customers. “Related to that is the opportunity to develop innovative and practical solutions for machine learning at scale, leveraging data coming to Alexa from the field,” he says. “For scientists, this is a great place for doing research because Amazon is very data-driven and metrics-oriented. The rigor expected in scientific publications in terms of running controlled experiments and tracking progress based on well-established metrics is exactly what we see in our team.”

“You also interact with people with strong technical and practical experience and learn from each other,” he says. “You’re exposed to a breadth of challenges, many of them long-term applications for artificial intelligence.”

Before coming to Amazon, Spyros conducted research in acoustic modeling for automatic speech recognition, speaker diarization, statistical machine translation, and language identification. He has over 60 publications in peer-reviewed conferences and journals, with three best paper awards. He earned a Master’s Degree in Computer Science from Northeastern University.

Outside of work, Spyros enjoys watching movies, playing the guitar, and traveling with his wife and three kids.