Developing Machine Learning-Driven Customer-Facing Product Features
Developing Machine Learning-Driven Customer-Facing Product Features


As Machine Learning becomes a core component of any forward-looking company, how can we weave ML-driven functionality into the products and services we offer? This talk will explain the methodology we follow at Square to develop ML-driven customer-facing product features, which is based on paying close attention to four key and interdependent aspects: Design, Modeling, Engineering, and Analytics. Design is concerned about the usefulness and remarkability of the feature, and thus cares about the overall functionality, ease of use, and aesthetics of the experience. Modeling is concerned about the accuracy of the ML model, and thus cares about the training data, the features and performance of the model, and —crucially for a customer-facing product— how the application behaves in the face of the mistakes the model will inevitably make (false positives, false negatives, lack of predictions above a certain confidence). Engineering in turn is concerned about running the ML model at scale, and thus cares about the latency, throughput, and robustness of the inferencing service. Finally, Analytics is concerned about the adoption of the feature, and thus cares about the instrumentation to capture detailed usage, the definition of success metrics and dashboards, and the collection of feedback in a manner that the ML model can learn from, and thus keep improving over time. When all these aspects align, we can create remarkable ML-powered experiences that delight our customers.


Marsal Gavalda is a senior R&D executive with deep expertise in speech, language, and machine learning technologies. Marsal currently leads the Commerce Platform Machine Learning team at Square, where he applies machine learning for economic empowerment and financial inclusion. Previously, Marsal headed the Machine Intelligence team at Yik Yak, where he developed natural language processing and machine learning services to analyze the content of messages, discover trends, and make recommendations at scale and across languages. Prior to that, Marsal served as the Director of Research at MindMeld (acquired by Cisco), where he applied the latest advances in speech recognition, language understanding, information retrieval, and machine learning to the MindMeld conversational and anticipatory computing platform. Marsal has also extensive experience in the customer interaction and speech analytics space, as he has served as VP and Chief of Research at Verint Systems and as VP of Research and Incubation at Nexidia (acquired by NICE), where he developed disruptive speech analytics solutions for the call center, intelligence, and media markets. Marsal holds a PhD in Language Technologies and a MS in Computational Linguistics, both from Carnegie Mellon University, and a BS in Computer Science from BarcelonaTech. Marsal is the author of over thirty technical and literary publications, thirteen issued patents, and is fluent in six languages. He is also a frequent speaker at academic and industry conferences and organizes, every summer, a science and humanities summit in Barcelona on topics as diverse as machine translation, music, or the neuroscience of free will.

Privacy Settings
We use cookies to enhance your experience while using our website. If you are using our Services via a browser you can restrict, block or remove cookies through your web browser settings. We also use content and scripts from third parties that may use tracking technologies. You can selectively provide your consent below to allow such third party embeds. For complete information about the cookies we use, data we collect and how we process them, please check our Privacy Policy
Consent to display content from - Youtube
Consent to display content from - Vimeo
Google Maps
Consent to display content from - Google