Sr. Software Development Engineer - Alexa AI Amazon
Seattle, WA

Job Description

Interested in Machine Learning? How about Natural Language Understanding (NLU)? Does it excite you to implement and scale scientific research at Amazon's scale? If so, you are looking at that right opening you have been searching for!

Since its initial launch, Alexa has been rapidly expanding with 100s of millions of echo devices being sold in every retail peak sale event. Amazon believes speech to be a pervasive and a superior mode of interacting with automated systems, and is committed to maintaining our industry-leading position in this space. Alexa AI's mission is to build super-scalable ML services to empower our customers with the magic of machine learning in innovative applications to solve their day-to-day pain points. Our team works closely with some of the best Applied ML Research Scientists and Language Engineers in the field. We realize their researches by building distributed systems that optimize these machine learning algorithms for production scale in the field of NLU. Once speech recognition converts speech to text, we take it from there and make sense of it. We are working hard and architecting the foundations of sustainable and scalable solutions on the cloud using AWS.

Under Alexa AI, our team is the de facto of all evaluations in NLU. From a product perspective, if research scientists are the Tony Starks we are the team that builds J.A.R.V.I.S. to help them understand what they build. Our services cater to evaluation use cases of NLU scientists by interpreting utterances with the NLU Engine, creating metrics, and comparing the metrics with benchmarks to determine if the model passed the evaluation test designed. For the team's technical secret sauce, our charter is to build any distributed infrastructure around NLU Engine to solve any pain point of NLU modeling. When you ask Alexa something, its 1 utterance at a time. However, when it gets to the throughput during modeling, we are talking 10 digits here! For example, before any change to Alexa's modeling logic goes live to our customers, we evaluate billions of utterances to insist on the highest standards.

When we talk of evaluations of models built using the teenager regressions algorithms; instantly AUCs, p-values & RMSEs come to our minds. Web searches reveal how research in this field is extremely nascent for NLP. Ever thought what these evaluation mechanisms would be for Alexa NLU? Join us to find out and make history in shaping the future of language processing!!