Interspeech
This year's Interspeech will be held in Graz, Austria, whose famed clock tower was built in the mid-1500s
Photo courtesy of Getty Images

The 16 Alexa-Related Papers at This Year’s Interspeech

At next week’s Interspeech, the largest conference on the science and technology of spoken-language processing, Alexa researchers have 16 papers, which span the five core areas of Alexa functionality: device activation, or recognizing speech intended for Alexa and other audio events that require processing; automatic speech recognition (ASR), or converting the speech signal into text; natural-language understanding, or determining the meaning of customer utterances; dialogue management, or handling multiturn conversational exchanges; and text-to-speech, or generating natural-sounding synthetic speech to convey Alexa’s responses. Two of the papers are also more-general explorations of topics in machine learning.

Device Activation

Model Compression on Acoustic Event Detection with Quantized Distillation
Bowen Shi, Ming Sun, Chieh-Chi Kao, Viktor Rozgic, Spyros Matsoukas, Chao Wang

The researchers combine two techniques to shrink neural networks trained to detect sounds by 88%, with no loss in accuracy. One technique, distillation, involves using a large, powerful model to train a leaner, more-efficient one. The other technique, quantization, involves using a fixed number of values to approximate a larger range of values.

Sub-band Convolutional Neural Networks for Small-footprint Spoken Term Classification
Chieh-Chi Kao, Ming Sun, Yixin Gao, Shiv Vitaladevuni, Chao Wang

Convolutional neural nets (CNNs) were originally designed to look for the same patterns in every block of pixels in a digital image. But they can also be applied to acoustic signals, which can be represented as two-dimensional mappings of time against frequency-based “features”. By restricting an audio-processing CNN’s search only to the feature ranges where a particular pattern is likely to occur, the researchers make it much more computationally efficient. This could make audio processing more practical for power-constrained devices.

A Study for Improving Device-Directed Speech Detection toward Frictionless Human-Machine Interaction
Che-Wei Huang, Roland Maas, Sri Harish Mallidi, Björn Hoffmeister

This paper is an update of prior work on detecting device-directed speech, or identifying utterances intended for Alexa. The researchers find that labeling dialogue turns (distinguishing initial utterances from subsequent utterances) and using signal representations based on Fourier transforms rather than mel-frequencies improve accuracy. They also find that, among the features extracted from speech recognizers that the system considers, confusion networks, which represent word probabilities at successive sentence positions, have the most predictive power.

Automatic Speech Recognition (ASR)

Acoustic Model Bootstrapping Using Semi-Supervised Learning
Langzhou Chen, Volker Leutnant

The researchers propose a method for selecting machine-labeled utterances for semi-supervised training of an acoustic model, the component of an ASR system that takes an acoustic signal as input. First, for each training sample, the system uses the existing acoustic model to identify the two most probable word-level interpretations of the signal at each position in the sentence. Then it finds examples in the training data that either support or contradict those probability estimates, which it uses to adjust the uncertainty of the ASR output. Samples that yield significant reductions in uncertainty are preferentially selected for training.

Improving ASR Confidence Scores for Alexa Using Acoustic and Hypothesis Embeddings
Prakhar Swarup, Roland Maas, Sri Garimella, Sri Harish Mallidi, Björn Hoffmeister

Speech recognizers assign probabilities to different interpretations of acoustic signals, and these probabilities can serve as inputs to a machine learning model that assesses the recognizer’s confidence in its classifications. The resulting confidence scores can be useful to other applications, such as systems that select machine-labeled training data for semi-supervised learning. The researchers append embeddings — fixed-length vector representations — of both the raw acoustic input and the speech recognizer’s best estimate of the word sequence to the inputs to a confidence-scoring network. The result: a 6.5% reduction in equal-error rate (the error rate that results when the false-negative and false-positive rates are set as equal).

Multi-Dialect Acoustic Modeling Using Phone Mapping and Online I-Vectors
Harish Arsikere, Ashtosh Sapru, Sri Garimella

Multi-dialect acoustic models, which help convert multi-dialect speech signals to words, are typically neural networks trained on pooled multi-dialect data, with separate output layers for each dialect. The researchers show that mapping the phones — the smallest phonetic units of speech — of each dialect to those of the others offers comparable results with shorter training times and better parameter sharing. They also show that recognition accuracy can be improved by adapting multi-dialect acoustic models, on the fly, to a target speaker.

Neural Machine Translation for Multilingual Grapheme-to-Phoneme Conversion
Alex Sokolov, Tracy Rohlin, Ariya Rastrow

Grapheme-to-phoneme models, which translate written words into their phonetic equivalents (“echo” to “E k oU”), enable speech recognizers to handle words they haven’t seen before. The researchers train a single neural model to handle grapheme-to-phoneme conversion in 18 languages. The results are comparable to those of state-of-the-art single-language models for languages with abundant training data and better for languages with sparse data. Multilingual models are more flexible and easier to maintain in production environments.

Scalable Multi Corpora Neural Language Models for ASR
Anirudh Raju, Denis Filimonov, Gautam Tiwari, Guitang Lan, Ariya Rastrow

Language models, which compute the probability of a given sequence of words, help distinguish between different interpretations of speech signals. Neural language models promise greater accuracy than existing models, but they’re difficult to incorporate into real-time speech recognition systems. The researchers describe several techniques to make neural language models practical, from a technique for weighting training samples from out-of-domain data sets to noise contrastive estimation, which turns the calculation of massive probability distributions into simple binary decisions.

Natural-Language Understanding

Neural Named Entity Recognition from Subword Units
Abdalghani Abujabal, Judith Gaspers

Named-entity recognition is crucial to voice-controlled systems — as when you tell Alexa “Play ‘Spirit’ by Beyoncé”. A neural network that recognizes named entities typically has dedicated input channels for every word in its vocabulary. This has two drawbacks: (1) the network grows extremely large, which makes it slower and more memory intensive, and (2) it has trouble handling unfamiliar words. The researchers trained a named-entity recognizer that instead takes subword units — characters, phonemes, and bytes — as inputs. It offers comparable performance with a vocabulary of only 332 subwords, versus 74,000-odd words.

Dialogue Management

HyST: A Hybrid Approach for Flexible and Accurate Dialogue State Tracking
Rahul Goel, Shachi Paul, Dilek Hakkani-Tür

Dialogue-based computer systems need to track “slots” — types of entities mentioned in conversation, such as movie names — and their values — such as Avengers: Endgame. Training a machine learning system to decide whether to pull candidate slot values from prior conversation or compute a distribution over all possible slot values improves slot-tracking accuracy by 24% over the best-performing previous system.

Towards Universal Dialogue Act Tagging for Task-Oriented Dialogues
Shachi Paul, Rahul Goel, Dilek Hakkani-Tür

Dialogue-based computer systems typically classify utterances by “dialogue act” — such as requesting, informing, and denying — as a way of gauging progress toward a conversational goal. As a first step in developing a system that will automatically label dialogue acts in human-human conversations (to, in turn, train a dialogue-act classifier), the researchers create a “universal tagging scheme” for dialogue acts. They use this scheme to reconcile the disparate tags used in different data sets.

Topical-Chat: Towards Knowledge-Grounded Open-Domain Conversations
Karthik Gopalakrishnan, Behnam Hedayatnia, Qinlang Chen, Anna Gottardi, Sanjeev Kwatra, Anu Venkatesh, Raefer Gabriel, Dilek Hakkani-Tür

The researchers report a new data set, which grew out of the Alexa Prize competition and is intended to advance research on AI agents that engage in social conversations. Pairs of workers recruited through Mechanical Turk were given information on topics that arose frequently during Alexa Prize interactions and asked to converse about them, documenting the sources of their factual assertions. The researchers used the resulting data set to train a knowledge-grounded response generation network, and they report automated and human evaluation results as state-of-the-art baselines.

Text-to-Speech

Towards Achieving Robust Universal Neural Vocoding
Jaime Lorenzo Trueba, Thomas Drugman, Javier Latorre, Thomas Merritt, Bartosz Putrycz, Roberto Barra-Chicote, Alexis Moinet, Vatsal Aggarwal

A vocoder is the component of a speech synthesizer that takes the frequency-spectrum snapshots generated by other components and fills in the information necessary to convert them to audio. The researchers trained a neural-network-based vocoder using data from 74 speakers of both genders in 17 languages. The resulting “universal vocoder” outperformed speaker-specific vocoders, even on speakers and languages it had never encountered before and unusual tasks such as synthesized singing.

Fine-Grained Robust Prosody Transfer for Single-Speaker Neural Text-to-Speech
Viacheslav Klimkov, Srikanth Ronanki, Jonas Rohnke, Thomas Drugman

The researchers present a new technique for transferring prosody (intonation, stress, and rhythm) from a recording to a synthesized voice, enabling the user to choose whose voice will read recorded content, with inflections preserved. Where earlier prosody transfer systems used spectrograms — frequency spectrum snapshots — as inputs, the researchers’ system uses easily normalized prosodic features extracted from the raw audio.

Machine Learning

Two Tiered Distributed Training Algorithm for Acoustic Modeling
Pranav Ladkat, Oleg Rybakov, Radhika Arava, Sree Hari Krishnan Parthasarathi,I-Fan Chen, Nikko Strom

When neural networks are trained on large data sets, the training needs to be distributed, or broken up across multiple processors. A novel combination of two state-of-the-art distributed-learning algorithms — GTC and BMUF — achieves both higher accuracy and more-efficient training then either, when learning is distributed to 128 parallel processors.

BMUF-GTC.gif._CB436386414_.gif
The researchers' new method splits distributed processors into groups, and within each group, the processors use the highly accurate GTC method to synchronize their models. At regular intervals, designated representatives from all the groups use a different method — BMUF — to share their models and update them accordingly. Finally, each representative broadcasts its updated model to the rest of its group.
Animation by Nick Little

One-vs-All Models for Asynchronous Training: An Empirical Analysis
Rahul Gupta, Aman Alok, Shankar Ananthakrishnan

A neural network can be trained to perform multiple classifications at once: it might recognize multiple objects in an image, or assign multiple topic categories to a single news article. An alternative is to train a separate “one-versus-all” (OVA) classifier for each category, which classifies data as either in the category or out of it. The advantage of this approach is that each OVA classifier can be re-trained separately as new data becomes available. The researchers present a new metric that enables comparison of multiclass and OVA strategies, to help data scientists determine which is more useful for a given application.

About the Author
Larry Hardesty is a science writer at Amazon. Previously, he was managing editor of the Boston Book Review, a senior editor at MIT Technology Review, and the computer science writer at the MIT News Office.

Work with us

See More Jobs
US, WA, Seattle
Amazon's Sponsored Ads is one of the fastest growing business domains and we are looking senior level scientists with experience building complex machine learning models and writing elegant prototype code. We are still in Day 1 and there is an abundance of opportunities that are yet to be explored. We are a team of highly motivated and collaborative applied scientists and software development engineers with an entrepreneurial spirit and bias for action. With the team growing at an unprecedented rate, there is broad mandate to experiment and innovate!Our systems and algorithms operate on one of the world's largest product catalogs, matching shoppers with products - with a high relevance bar and strict latency constraints. The ads sourcing works across the spectrum of ad serving including coverage expansion, increasing utilization of tail detail pages, ad relevance, ad quality, collaborative filtering, and much more. Our technology enables thousands of brands, sellers and authors to drive discovery and sales of their products at Amazon by millions of customers.We are looking for an Applied Scientists, with a background in Machine Learning to optimize serving ads on billions of product pages. The solutions you create/deploy would ensure relevant ads are served to Amazon's customers. You will directly impact the shopping experience while helping our advertisers get the maximum ROI. You will be expected to demonstrate strong ownership and should be curious to learn and leverage the rich textual, image, and other contextual signals.This role will challenge you to utilize cutting-edge machine learning techniques in the domain of predictive modeling, natural language processing (NLP), deep learning, and image recognition to deliver significant impact for the business. Ideal candidates will be able to work cross functionally across multiple stakeholders, synthesize the science needs of our business partners, develop models to solve business needs, and implement solutions in production.
US, NY, New York
Amazon’s Advertising Analytics and Insights (A&I) team is looking for a motivated and experienced Applied Scientist to pioneer new ways to analyze the needs, experiences, and behaviors of Amazon consumers at petabyte scale. We are a "startup" within one of the fastest growing and most strategic parts of Amazon. Analytics & Insights owns the product, technology and deployment roadmap for advanced analytics and insights products across our advertiser success team. Advertiser success is core to Amazon’s growth, as it helps our suppliers drive awareness, consideration, and purchase of their products by hundreds of millions of consumers around the world, and generates revenue which helps us lower prices and invest in improvements to our customer experience. We are a highly motivated, collaborative and fun-loving team with an entrepreneurial spirit and bias for action. With a broad mandate to experiment and innovate, we are growing at an unprecedented rate with a seemingly endless range of new opportunities.Marketers are looking for ways to better understand consumer needs and behaviors. Understanding the “consumer journey” helps marketers cater to consumer needs and optimize their advertising spending, increasing the value of advertising for both advertisers and consumers. In this role, you will take a leading role in defining the problem and our approach to the solution. You will join a new and growing team of top-notch technical professionals working to develop and evaluate new advertising metrics based on petabytes of data. You will innovate on behalf of our customers by prototyping, delivering functional proofs of concept (POCs), and partnering with our engineers to productize and scale successful POCs. If you are passionate about creating the future, come join us as we work hard, have fun, and make history.
US, CA, Sunnyvale
Your machine learning skills can help make that a reality on the Amazon Music search team. We are seeking an experienced Machine Learning Scientist who will join a team of experts in the field of machine learning, and work together to break new ground in the world of understanding and classifying different forms of music, and creating interactive experiences to help users find the music they are in the mood for. We work on machine learning problems for music classification, recommender systems, dialogue systems, NLP, and music information retrieval.You'll work in a collaborative environment where you can pursue ambitious, long-term research, with many peta-bytes of data, work on problems that haven’t been solved before, quickly implement and deploy your algorithmic ideas at scale, understand whether they succeed via statistically relevant experiments across millions of customers, and publish your research. You'll see the work you do directly improve the experience of Amazon Music customers on Alexa/Echo, mobile, and web.Amazon MusicImagine being a part of an agile team where your ideas have the potential to reach millions. Picture working on cutting-edge consumer-facing products, where every single team member is a critical voice in the decision-making process. Envision being able to leverage the resources of a Fortune-500 company within the atmosphere of a start-up. Welcome to Amazon Music, where ideas are born and come to life as Amazon Music Unlimited, Prime Music, and so much more.Everyone on our team has a meaningful impact on product features, new directions in music streaming, and customer engagement. We are looking for new team members across a variety of job functions including software engineering/development, marketing, design, ops and more. Come join us as we make history by launching exciting new projects in the coming year.Our team is focused on building a personalized, curated, and seamless music experience. We want to help our customers discover up-and-coming artists, while also having access to their favorite established musicians. We build systems that are distributed on a large scale, spanning our music apps, web player, and voice-forward audio engagement on mobile and Amazon Echo devices, powered by Alexa to support our customer base. Amazon Music offerings are available in countries around the world, and our applications support our mission of delivering music to customers in new and exciting ways that enhance their day-to-day lives.Come innovate with the Amazon Music team!
DE, BE, Berlin
Amazon is looking for creative Applied Scientists to tackle some of the most interesting problems on the leading edge of Machine Learning (ML), Natural Language Processing (NLP), Question Answering (QA), and Information Retrieval (IR) with our Alexa search team. Alexa search is part of our ongoing Alexa Information efforts focused on reinventing information extraction and retrieval for a voice-forward, multi-modal future.The successful candidate will develop novel ML/NLP/QA/IR/Deep Learning technologies to make Alexa smarter. They will have a true passion for working in a collaborative, cross-functional environment that encourages thinking about optimized solutions to unique problems that do not yet have a known science solution.If you are looking for an opportunity to solve deep technical problems and build innovative solutions in a fast-paced environment working within a smart and passionate team, this might be the role for you. You will develop and implement novel algorithms and modeling techniques to leverage and advance the state-of-the-art in technology areas that are found at the intersection of ML, NLP, IR, and Deep Learning. Your work will directly impact Amazon products and services that make use of speech and language technology. You will gain hands on experience with Alexa and large-scale computing resources.In this role you will:· Work collaboratively with scientists and developers to design and implement automated, scalable NLP/ML/QA/IR models for accessing and presenting information;· Drive scalable solutions from the business, to prototyping, production testing and through engineering directly to production;· Drive best practices on the team, deal with ambiguity and competing objectives, and mentor and guide junior members to achieve their career growth potential.
DE, BE, Berlin
Amazon is looking for creative Applied Scientists to tackle some of the most interesting problems on the leading edge of Machine Learning (ML), Natural Language Processing (NLP), Question Answering (QA), and Information Retrieval (IR) with our Alexa search team. Alexa search is part of our ongoing Alexa Information efforts focused on reinventing information extraction and retrieval for a voice-forward, multi-modal future.The successful candidate will develop novel ML/NLP/QA/IR/Deep Learning technologies to make Alexa smarter. They will have a true passion for working in a collaborative, cross-functional environment that encourages thinking about optimized solutions to unique problems that do not yet have a known science solution.If you are looking for an opportunity to solve deep technical problems and build innovative solutions in a fast-paced environment working within a smart and passionate team, this might be the role for you. You will develop and implement novel algorithms and modeling techniques to leverage and advance the state-of-the-art in technology areas that are found at the intersection of ML, NLP, IR, and Deep Learning. Your work will directly impact Amazon products and services that make use of speech and language technology. You will gain hands on experience with Alexa and large-scale computing resources.In this role you will:· Work collaboratively with scientists and developers to design and implement automated, scalable NLP/ML/QA/IR models for accessing and presenting information;· Drive scalable solutions from the business, to prototyping, production testing and through engineering directly to production;· Drive best practices on the team, deal with ambiguity and competing objectives, and mentor and guide junior members to achieve their career growth potential.
DE, BE, Berlin
Amazon's Intelligent Cloud Control Group in Berlin is looking for an Applied Science Manager to lead the research and development of large-scale machine learning implementations that will revolutionize the way Amazon operates the tens of thousands of software services and subcomponents across our retail businesses. With an ever-growing number of fleets, developers, customers, products, marketplaces, sellers, and businesses, the Amazon service graph is one of the largest and most complex tech ecosystems in the world. We are building an Intelligent Cloud Control system that enables Amazon businesses (Retail, Amazon Video, Kindle, and more) to accelerate innovation in the cloud.As an Applied Science Manager in the Intelligent Cloud Control team, you'll leverage your own skills and those of your team of machine learning engineers, data scientists, and applied scientists to develop and evaluate machine learning models using extremely large datasets such as the orders, website traffic, telemetry, and logs from every host at Amazon. Our datasets extend into the multi-exabyte range, and our science products are of critical importance to the retail businesses of Amazon. You will own researching, developing, prioritizing, and releasing both prototypes and reliable automated production workflow for the model. You will collaborate with other managers and leaders to improve the Amazon retail customer experience.Our responsibility is to maximize the availability and contribute to the better efficiency of Amazon’s retail experience, so our opportunities are endless. From natural language processing and information extraction of operational issues to unsupervised multi-variate anomaly detection to discover nodes of linked sub-system behavior, the insights and opportunities to discover and remediate customer-impacting issues are profound and our solutions worthy of publication.We’re looking for engineers capable of using machine learning and other techniques to design, evaluate, and implement state-of-the-art solutions for never-before-solved problems.
US, NY, New York
Amazon is building a world class advertising business and defining and delivering a collection of self-service performance advertising products that drive discovery and sales of merchandise. Our products are strategically important to our Retail and Marketplace businesses driving long term growth. We deliver billions of ad impressions and millions of clicks daily and are breaking fresh ground to create world-class products.Our team's mandate is to accelerate advertising in the Books category as it requires a differentiated customer and advertiser experience. We own end-to-end the advertising experience including placements, ad relevance, creative, ad serving, advertiser experience, and marketing. We are looking for entrepreneurial, innovative individuals who thrive on solving tough problems. We are investing in a deep science and technical team in the heart of NYC to pursue this exciting opportunity.About this role:We are seeking a talented, energetic, entrepreneurial, and self-driven Applied Scientist with a background in Machine Learning to join our team. In this role, you will optimize serving ads on billions of product pages. The solutions you create would drive step increases in coverage of search and brand ads across the retail website and ensure relevant ads are served to Amazon's customers. You will directly impact our customers’ shopping experience while helping our sellers get the maximum ROI from advertising on Amazon. You will be expected to demonstrate strong ownership and should be curious to learn and leverage the rich textual, image, and other contextual signals.This role will challenge you to utilize cutting-edge machine learning techniques in the domain of predictive modeling, natural language processing (NLP), deep learning, reinforcement learning, query understanding, and vector search (kNN) to deliver significant impact for the business. Ideal candidates will be able to work cross functionally across multiple stakeholders, synthesize the science needs of our business partners, develop models to solve business needs, and implement solutions in production.
IN, TN, Chennai
Amazon Device software team is looking for an exceptional, motivated Applied Scientist to implement inference of Deep learning models on the Edge. This is an ambitious undertaking to create a new device category from ground up and is replete with technical challenges. You will take an exciting and challenging role on a team and work with world-class engineers, business leaders, and other talented engineering teams while you design, architect, and code for our Deep learning inference platform for the edge.You will be an owner for features, including design, development, testing, and operations. You'll also work closely with engineers from other Amazon teams. This is a great opportunity for you to develop new algorithms and to innovate in this space.This position requires the ability to work on inference scheduling and concurrency on resource constrained computing platforms on the edge, analyze performance of inference implementations, knowledge of GPU/accelerator architecture and pit falls, minimal working knowledge of GPU programming (CUDA or OpenCL).From a soft skills standpoint, you will collaborate across Amazon to influence and be influenced regarding best practices for machine learning inference. Additionally, you'll need to possess a passion for artificial intelligence and computer vision, with knowledge of latest developments in DL and AI.
US, NY, New York
Are you looking to solve exciting Fraud challenges in the cloud? Do you want to work on a Cloud business that is growing multiple fold annually? If so, we want to talk to you! We are seeking a talented Data Scientist to help us evolve our processes and to improve customer experience across the globe.The mission of the Fraud Prevention Team is to keep the AWS platform a safe and trusted place for our customers and partners. We achieve this goal by identifying and preventing fraud and abuse for all AWS services, worldwide. Every day fraudsters attempt to steal our services. We use extensive data and machine learning models to build software solutions to detect and prevent such activities.As a Data Scientist, you will work directly with Research Scientists and Software Development Engineers to monitor the flavor/ trend of fraud worldwide and create appropriate actions to respond in a collaborative environment. There are no walls, and success is determined by your ability to dive deep, and understand the subtle demands new and complex services will place upon systems and teams.What would I do?· · Lead day-to-day review of emerging fraud threats and work with the broader team to drive responses to combat them· Manage your own process: identify and execute on high impact projects, triage external requests, and make sure you bring projects to conclusion in time for the results to be useful· Deep dive on the problems using SQL and scripting language like R, Python and internal dashboards and drive short term and long term solutions· Analyze data (past customer behavior, sales inputs, and other sources) to figure out trends, and output reports with clear recommendations.· Collaborate closely with the development team to recommend innovations based on data analysis· · Handle escalations from AWS Management for Fraud and related activities· · Be the primary point of contact during fraud outbreaks – including analyses to identify responses and applying responses.· Apply Statistical and Machine Learning methods to specific business problems and data· · Drive the design and development of tools to aid in operations and automation, such as creation of right dashboards to improve monitoring fraud· · Collaborate in a fast paced environment with multiple teams and customers in a dynamic entrepreneurial organization
UK, Cambridge
“Alexa find me a job on the Alexa Knowledge team!”Our focus in the Alexa Knowledge team combines natural language understanding, acquiring large volumes of structured knowledge, and building autonomous machine reasoning to allow our customers to get answers to their questions in the most natural way possible. We’re part of a huge research and engineering effort on the Amazon Alexa team.We’ve solved many complex problems to get to where we are today, but there are still plenty of challenges ahead of us, and Alexa is getting smarter every day. The problems we solve in the Alexa Knowledge team in Cambridge help Alexa get smarter by understanding the different ways people talk, by learning more and more facts about the world, by improving her common sense reasoning and by responding in the most natural way possible in multiple languages. We set out to build Alexa at Amazon because we believe that voice will fundamentally improve the way people will interact with technology, and we wanted to create a computer that could be controlled entirely by your voice.On the Alexa Knowledge team, we are constantly making Alexa smarter by enabling her to learn about what’s going on in the world. We use many different techniques to enable learning and reasoning across a range of structured and unstructured data. We are constantly at the forefront of both research and engineering in understanding user demands and data sources, to extract the right knowledge, expand the range of, and how, Alexa accesses information – all to improve Alexa and give users the best experience. We believe that the information to answer (almost) every question can be found somewhere on the internet, and not just in an encyclopedia. Our goal is to teach Alexa how to autonomously consume a wider range of texts and expand her Knowledge Graph to learn more about the world.We use multiple Machine Learning and Natural Language Processing techniques to enable learning and reasoning across a range of structured data sources, at internet-level scale. This requires methods that lie beyond the cutting edge academic and industrial research of today. The scope of this role is broad, covering a diverse range of problem space that include, but are not limited to ontology expansion and alignment techniques, fact extraction from structured text, large-scale categorisation, supervised, weakly supervised or self-supervised learning methods etc. As a senior scientist, you will bring academic and/or industrial practical experience and create novel solutions to complex problems. You will guide junior scientists, collaborate with the best researchers in the field and work with the engineering team to bring your solutions to the millions of customers who use Alexa every day.
UK, Cambridge
“Alexa find me a job on the Alexa Knowledge team!”Our focus in the Alexa Knowledge team combines natural language understanding, acquiring large volumes of structured knowledge, and building autonomous machine reasoning to allow our customers to get answers to their questions in the most natural way possible. We’re part of a huge research and engineering effort on the Amazon Alexa team.We’ve solved many complex problems to get to where we are today, but there are still plenty of challenges ahead of us, and Alexa is getting smarter every day. The problems we solve in the Alexa Knowledge team in Cambridge help Alexa get smarter by understanding the different ways people talk, by learning more and more facts about the world, by improving her common sense reasoning and by responding in the most natural way possible in multiple languages. We set out to build Alexa at Amazon because we believe that voice will fundamentally improve the way people will interact with technology, and we wanted to create a computer that could be controlled entirely by your voice.On the Alexa Knowledge team, we are constantly making Alexa smarter by enabling her to learn about what’s going on in the world. We use many different techniques to enable learning and reasoning across a range of structured and unstructured data. We are constantly at the forefront of both research and engineering in understanding user demands and data sources, to extract the right knowledge, expand the range of, and how, Alexa accesses information – all to improve Alexa and give users the best experience. We believe that the information to answer (almost) every question can be found somewhere on the internet, and not just in an encyclopedia. Our goal is to teach Alexa how to autonomously consume a wider range of texts and expand her Knowledge Graph to learn more about the world.We use multiple Machine Learning and Natural Language Processing techniques to validate the information extracted from various structured and unstructured sources, at internet-level scale. This requires methods that lie beyond the cutting edge academic and industrial research of today. The scope of this role is broad, covering a diverse range of problem space that include, but are not limited to fact verification, natural language inference, large-scale categorisation, weakly supervised or self-supervised learning methods etc. As a senior scientist, you will bring academic and/or industrial practical experience and create novel solutions to complex problems. You will guide junior scientists, collaborate with the best researchers in the field and work with the engineering team to bring your solutions to the millions of customers who use Alexa every day.
US, TX, Austin
Have you ever ordered a product on Amazon and when that box with the smile arrives you wonder how it got to you so fast? Wondered where it came from and how much it would have cost Amazon? If so, Amazon’s Supply Chain Optimization Technologies (SCOT) team is for you.We build systems to peer into the future and estimate the distribution of tens of millions of products every week to Amazon’s warehouses in the most cost-effective way. When customers place orders, our systems use real time, large scale optimization techniques to optimally choose where to ship from and how to consolidate multiple orders so that customers get their shipments on time or faster with the lowest possible transportation costs. This team is focused on saving hundreds of millions of dollars using cutting edge science, machine learning, and scalable distributed software on the Cloud that automates and optimizes inventory and shipments to customers under the uncertainty of demand, pricing and supply.Watch this video to learn more about our organization, SCOT: http://bit.ly/amazon-scotIn S&OP we use cutting-edge machine learning and scalable distributed software in the Cloud to predict flows of products between our warehouses world-wide and distribute tens of millions of products every week in the most cost-effective way. We drive towards end to end automation solution requiring the use of machine learning in a wide variety of ways to forecasting, data anomaly detection, measuring the impact of our forecasts and bringing back this impact back to complete a closed loop process. You’ll have a WW impact in working with teams around the world in solving the unique challenges in each country that Amazon operates in.The Data Scientist, in partnership with the Product Management, Operations, and Tech teams will lead efforts in three areas:1) Building models to forecast inbound and outbound unit volumes through the world2) Identifying opportunities for forecast optimization by working with teams downstream to measure the results of or forecasts3) Evolve our anomaly detection efforts to ensure our automation efforts produce high quality results and issues are caught before the impact FC operations.Amazon is an Equal Opportunity-Affirmative Action Employer – Minority / Female / Disability / Veteran / Gender Identity / Sexual Orientation / Age
US, CA, San Francisco
Can Alexa help anyone experience the music they enjoy? Even if they don't know what they'd like to listen to in this moment? Or, if they know they want “Happy rock from the 90s”, can she help them find it?Your machine learning skills can help make that a reality on the Amazon Music team. We are seeking an Applied Scientist who will join a team of experts in the field of machine learning, and work together to break new ground in the world of understanding and classifying different forms of music, and creating interactive experiences to help users find the music they are in the mood for. We work on machine learning problems for music classification, conversational AI, NLP, and music information retrieval.You'll work in a collaborative environment where you can pursue ambitious, long-term research, with many peta-bytes of data, work on problems that haven’t been solved before, quickly implement and deploy your algorithmic ideas at scale, understand whether they succeed via statistically relevant experiments across millions of customers, and publish your research. You'll see the work you do directly improve the experience of Amazon Music customers on Alexa/Echo, mobile, and web.The successful candidate will have a PhD in Computer Science with a strong focus on machine learning, or a related field, and 2+ years of practical experience applying ML to solve complex problems in signal processing, NLP or dialogue systems. Great if you have a passion for music, but this is not a requirement.Responsibilities:- Advance long-term, exploratory research projects in machine learning and related fields to create highly innovative customer experiences;- Analyze large amounts of Amazon’s customer data to discover patterns, find opportunities, and develop highly innovative, scalable algorithms to seize these opportunities;- Validate new or improved models via statistically relevant experiments across millions of customers;- Work closely with software engineering teams to build scalable prototypes for testing, and integrate successful models and algorithms in production systems at very large scale.Amazon MusicImagine being a part of an agile team where your ideas have the potential to reach millions. Picture working on cutting-edge consumer-facing products, where every single team member is a critical voice in the decision-making process. Envision being able to leverage the resources of a Fortune-500 company within the atmosphere of a start-up. Welcome to Amazon Music, where ideas are born and come to life as Amazon Music Unlimited, Prime Music, and so much more.Everyone on our team has a meaningful impact on product features, new directions in music streaming, and customer engagement. We are looking for new team members across a variety of job functions including software engineering/development, marketing, design, ops and more. Come join us as we make history by launching exciting new projects in the coming year.Our team is focused on building a personalized, curated, and seamless music experience. We want to help our customers discover up-and-coming artists, while also having access to their favorite established musicians. We build systems that are distributed on a large scale, spanning our music apps, web player, and voice-forward audio engagement on mobile and Amazon Echo devices, powered by Alexa to support our customer base. Amazon Music offerings are available in countries around the world, and our applications support our mission of delivering music to customers in new and exciting ways that enhance their day-to-day lives.Come innovate with the Amazon Music team!
CA, ON, Toronto
Have you ever ordered a product on Amazon and when that box with the smile arrives you wonder how it got to you so fast? Wondered where it came from and how much it would have cost Amazon? If so, Amazon’s Supply Chain Optimization Technologies (SCOT) team is for you. We build systems to peer into the future and estimate the distribution of tens of millions of products every week to Amazon’s warehouses in the most cost-effective way. When customers place orders, our systems use real time, large scale optimization techniques to optimally choose where to ship from and how to consolidate multiple orders so that customers get their shipments on time or faster with the lowest possible transportation costs. This team is focused on saving hundreds of millions of dollars using cutting edge science, machine learning, and scalable distributed software on the Cloud that automates and optimizes inventory and shipments to customers under the uncertainty of demand, pricing and supply.Fulfillment-by-Amazon (FBA) Inventory Optimization (FIO) is a relatively new team at Amazon’s Supply Chain Optimization Technologies (SCOT). We focus on driving long term free cash flow by automating and optimizing our third-party supply chain. The team’s efforts will address the key challenges facing the worldwide FBA Seller business, including 1) improving FBA Seller inventory efficiency, 2) efficiently balancing the supply and demand of FBA Seller capacity, 3) closing worldwide selection gap by enabling global selling profitability, and 4) driving out costs across the FBA supply chain to spin the flywheel. This is truly a unique problem space – optimizing for inventory in Amazon’s pipeline when you don’t control the process or own the inventory.FIO is seeking an Research Scientist to join its cross-functional team of data, applied and research scientists, economists, engineers, and product managers to utilize cutting edge optimization models, econometrics, machine-learning, and distributed software on the Cloud to build systems that automate and optimize inventory management under the uncertainty of demand, pricing and supply. We are recruiting a curious and creative Research Scientist who will collaborate with other scientists and engineers to leverage new machine learning methods and algorithms for the modeling and analysis of data.
US, WA, Seattle
At Amazon, we're working to be the most customer-centric company on earth. To get there, we need exceptionally talented, bright, and driven people. Amazon is seeking a Sr Simulation and Innovation Engineer specialization in discrete event simulation and optimization of material process flow of complete Warehouse operations for our World Wide Design & Innovation Engineering Team. Amazon is one of the most recognizable brand names in the world and we distribute millions of products each year to our loyal customers.Responsibilities:· Lead system level complex Discrete Event Simulation (DES) projects to build , simulate, and optimize the fulfillment center operational process flow models using FlexSim, Demo 3D or any other Discrete Event Simulation (DES) software packages· Understand process flows , analyze data, perform Design of Experiments and effectively represent in simulation model to achieve better correlation and process improvements· Perform DES process flow simulations to design, build, and improve order fulfillment infrastructure throughout the large-scale supply chain network.· Manage multiple DES simulation projects and tasks simultaneously and effectively influence, negotiate, and communicate with internal and external business partners, contractors and vendors.Facilitate process improvement initiatives among site operations, engineering, and corporate systems groups.· Provide technical leadership for large-scale industrial engineering projects MS Excel, AutoCAD, and MS Projects.· Work with complex MHE and process design and influence multiple teams working closely with business teams to build consensus among discordant views· Lead and coordinate simulation and design efforts between internal teams and outside vendors to develop optimal solutions for the network, including equipment specification, material flow, process design, and site layout.· Deliver results according to project schedules and quality· Provide leadership and coordination between internal departments and vendors for multiple sites.· Develop design solutions to the best-in-class process flow to improve the throughput of the fulfillment facilities· Make technical trade-offs for long term/short-term needs considering challenges in business area by applying relevant data science disciplines, and interactions among systems.
US, WA, Seattle
Are you passionate about developing new state-of-the-art measurement approaches at Petabyte scale? Amazon Advertising is one of Amazon’s fastest growing businesses, and we are leveraging our unique data, the latest machine learning methods and big data technologies to better understand how advertising influences customer behavior. We are looking for an Applied Scientist to develop new systems and methods in the most challenging and data rich areas of marketing. We need an expert in experimental statistics, machine learning or causal inference to design advanced new models with our world class data systems.As part of the 1PM team, this role will partner with a dedicated engineering team measuring the impact Amazon's advertising and identifying opportunities for optimization at scale. We drive initiatives to make smarter marketing decisions and improve the relevance of advertising to our customers. We move away from industry standard measurement systems and build sophisticated and insightful decision engines. We enable massive advertising programs, generating billions of impressions with decorated with rich representations of customer state. The major challenges we are solving include integrating petabyte-scale distributed retail systems into a singular service to synthesize e-commerce data into measurement and optimization models. The successful candidate will have a causal inference background, a start-up mentality, an appreciation for white-space, and success solving problems with large data sets.Key responsibilities include:· Scientists at Amazon are expected to develop new techniques to process large data sets and contribute to design of automated systems.· Apply ML, statistics or econometrics knowledge to develop and analyze prototype models.· Design and analyze data from large-scale online experiments in order to validate prototype models.· Collaborate with scientists across teams in peer-review processes, publishing research in internal forms and industry conferences.· Partner closely with product and engineering teams to develop new measurement systems and translate prototype models to production.· Establish scalable, efficient, and automated processes for large scale model development, validation, and implementation.· Research and experiment with novel statistical modeling approaches.
US, WA, Seattle
Looking for a career at a company that seeks to be Earth’s most customer-centric company? If so, why not consider joining Amazon!Are you passionate about leveraging data to deliver actionable insights that impacts the daily business decisions at Amazon? Does the prospect of dealing with large data excite you? Do you like getting "scrappy" with data to answer challenging product and customer behavior questions? Do you enjoy building flexible, performant, and global solutions for complex financial and risk problems? If so, here is a great opportunity to consider!Amazon Business Payments (ABP) is seeking an outstanding Data Scientist who combines their technical expertise with business intuition to build flexible, performant, and global solutions for complex financial and risk problems. You will generate critical insights to set the strategic direction to enhance our product features & processes that will delight our customers.As a Data Scientist on the ABP Research team, you will design and build systems that support financial products. You will work closely with business partners, software and data engineers to build scalable solutions that deliver exceptional value for our customers. You will utilize intellectual and technical capabilities, problem solving and analytical skills, and excellent communication to deliver customer value. You will partner with product and operations management to launch new, or improve existing, financial products within Amazon Business.Amazon Business Lending’s Pay By Invoice (PBI) is an exciting program designed to help businesses with lines of credit to facilitate their business spending needs. We take pride in building solutions that leverage key insights to help our business customers to be successful. We are looking for a talented and passionate Data Scientist who can develop a reliable, scalable, and technical infrastructure to serve our business customer needs and improve portfolio performance as we expand the program.ResponsibilitiesYou will help create our data assets, then use necessary technical methods and conduct analyses to derive insights that are critical to business success. You will be responsible for researching insights as well as educating the business, product, marketing, and business development teams on the implementation of those insights to enable data-driven, day-to-day decision making. You will partner with our marketing, product management, global engineering, operations and Finance teams to:· Contribute to the development and enhancement of business payment products and features.· Use data mining, model building, and other analytical techniques to develop and maintain customer segmentation and predictive models to drive the business and improve our machine learning engine.· Make recommendations for new metrics, techniques, and strategies to improve campaign targeting and measurement.· Improve targeting capabilities and uncover hidden opportunities using data, analytics and machine learning.· Understand business and product strategies, goals and objectives. Set the analytics roadmap to drive the goals of the business.· Own the analytics for one or more product areas, lead planning, execution and delivery of projects· Analyze and solve problems at their root, stepping back to understand the broader context.· Interface with all internal related and ancillary teams to deliver data and analytics as requested.· Provide support on experimental design, exploratory data analysis, and data management.
US, WA, Seattle
In the Amazon Selection Monitoring team, we want Amazon to have a complete awareness of all products on earth. We aggregate and identify all products along with complete and accurate facts. Our goal is to enrich and increase the coverage of Amazon product selection guided by consumers’ interests. We are establishing the most comprehensive, accurate and fresh universal selection of products.We have multiple position for applied scientists who are excited to work in big data challenges including; web scale data integration, entity and product matching, improving data quality, natural language processing, discovery of new relationships along with its semantic, knowledge inferencing and enhancement to support strategic and tactical decision-making.We are looking for applied scientists with experience in building practical solutions and can work closely with software engineers to ship and automate solutions in production. Our applied scientist also collaborate and partner with other teams across Amazon to understand and reflect on how to create benefit for our customer.Amazon is an Equal Opportunity-Affirmative Action Employer – Minority / Female / Disability / Veteran / Gender Identity / Sexual Orientation / Age.
US, WA, Seattle
In the Amazon Selection Monitoring team, we want Amazon to have a complete awareness of all products on earth. We aggregate and identify all products along with complete and accurate facts. Our goal is to enrich and increase the coverage of Amazon product selection guided by consumers’ interests. We are establishing the most comprehensive, accurate and fresh universal selection of products.We have multiple position for applied scientists who are excited to work in big data challenges including; web scale data integration, entity and product matching, improving data quality, natural language processing, discovery of new relationships along with its sematic, knowledge inferencing and enhancement to support strategic and tactical decision-making.We are looking for applied scientists with experience in building practical solutions and can work closely with software engineers to ship and automate solutions in production. Our applied scientist also collaborate and partner with other teams across Amazon to understand and reflect on how to create benefit for our customer.Amazon is an Equal Opportunity-Affirmative Action Employer – Minority / Female / Disability / Veteran / Gender Identity / Sexual Orientation / Age.
US, WA, Seattle
Amazon is looking for a passionate, talented, and inventive Scientist with a strong machine learning background to help build industry-leading Speech and Language technology.Our mission is to provide a delightful experience to Amazon’s customers by pushing the envelope in Automatic Speech Recognition (ASR), Natural Language Understanding (NLU), Machine Learning (ML), and Audio Signal Processing technologies.As part of our speech and language team, you will work alongside internationally recognized experts to develop novel algorithms and modeling techniques to advance the state-of-the-art in spoken language understanding. Your work will directly impact millions of our customers in the form of products and services that make use of speech and language technology. You will gain hands on experience with Amazon’s heterogeneous speech, text, and structured data sources, and large-scale computing resources to accelerate advances in spoken language understanding. It is not imperative to have experience in ASR. We have scientists building production models released to Echo customers, who had no prior speech experience, but very strong in ML, statistics, coding (and “can do” spirit!).We are hiring in all areas of spoken language understanding: ASR, NLU, text-to-speech (TTS), and Dialog Management.