Amazon’s new research on automatic speech recognition

Interspeech papers include novel approaches to speaker identification and the training of end-to-end speech recognition models.

As the largest conference devoted to speech technologies, Interspeech has long been a showcase for the latest research on automatic speech recognition (ASR) from Amazon Alexa. This year, Alexa researchers had 12 ASR papers accepted at the conference.

Diagram illustrating the architecture of the RNN-T ASR system.
The architecture of the RNN-T ASR system. Xt indicates the current frame of the acoustic signal. Yu-1 indicates the sequence of output subwords corresponding to the preceding frames.
From "Efficient minimum word error rate training of RNN-transducer for end-to-end speech recognition"

One of these, “Speaker identification for household scenarios with self-attention and adversarial training”, reports the speech team’s recent innovations in speaker ID, or recognizing which of several possible speakers is speaking at a given time.

Two others — “Subword regularization: an analysis of scalability and generalization for end-to-end automatic speech recognition” and “Efficient minimum word error rate training of RNN-transducer for end-to-end speech recognition” —examine ways to improve the quality of speech recognizers that use an architecture know as a recurrent neural network-transducer, or RNN-T.

In his keynote address this week at Interspeech, Alexa director of ASR Shehzad Mavawalla highlighted both of these areas — speaker ID and the use of RNN-Ts for ASR — as ones in which the Alexa science team has made rapid strides in recent years.

Speaker ID

Speaker ID systems — which enable voice agents to personalize content to particular customers — typically rely on either recurrent neural networks or convolutional neural networks, both of which are able to track consistencies in the speech signal over short spans of time. 

In “Speaker identification for household scenarios with self-attention and adversarial training”, Amazon applied scientist Ruirui Li and colleagues at Amazon, the University of California, Los Angeles, and the University of Notre Dame instead use an attention mechanism to identify longer-range consistencies in the speech signal.

In neural networks — such as speech processors — that receive sequential inputs, attention mechanisms determine which other elements of the sequence should influence the network’s judgment about the current element. 

Speech signals are typically divided into frames, which represent power concentrations at different sound frequencies over short spans of time. For a given utterance, Li and his colleagues’ model represents each frame as a weighted sum of itself and all the other frames in the utterance. The weights depend on correlations between the frequency characteristics of the frames; the greater the correlation, the greater the weight.

This representation has the advantage of capturing the distinctive properties of a speaker’s voice conveyed by each frame but suppressing accidental properties that are unique to individual frames and less characteristic of the speaker’s voice as a whole. 

These representations pass to a neural network that, during training, learns which of these properties are the best indicators of a speaker’s identity. Finally, the sequential outputs of this network — one for each frame — are averaged together to produce a snapshot of the utterance as a whole. These snapshots are compared to stored profiles to determine the speaker’s identity.

Li and his colleagues also used a few other tricks to make their system more reliable, such as adversarial training.

In tests, the researchers compared their system to four prior systems and found that its speaker identifications were more accurate across the board. Compared to the best-performing of the four baselines, the system reduced the identification error rate by about 12% on speakers whose utterances were included in the model training data and by about 30% on newly encountered speakers.

The RNN-T architecture

Another pair of papers examine ways to improve the quality of speech recognizers that use the increasingly popular recurrent-neural-network-transducer architecture, or RNN-T. An RNN-T processes a sequence of inputs in order, so that the output corresponding to each input factors in both the inputs and outputs that preceded it. 

Illustration of a series of possible subword segmentations of the speech input, with the probability of each.
A series of possible subword segmentations of the speech input, with the probability of each.
From “Subword regularization: an analysis of scalability and generalization for end-to-end automatic speech recognition”

In the ASR application, the RNN-T takes in frames of an acoustic speech signal and outputs text — a sequence of subwords, or word components. For instance, the output corresponding to the spoken word “subword” might be the subwords “sub” and “_word”. 

Training the model to output subwords keeps the network size small. It also enables the model to deal with unfamiliar inputs, which it may be able to break into familiar components.

In the RNN-T architecture we consider, the input at time t — the current frame of the input speech — passes to an encoder network, which extracts acoustic features useful for speech recognition. At the same time, the current, incomplete sequence of output subwords passes to a prediction network, whose output indicates likely semantic properties of the next subword in the sequence.

These two representations — the encoding of the current frame and the likely semantic properties of the next subword — pass to another network, which on the basis of both representations determines the next word in the output sequence.

New wrinkles

Subword regularization: an analysis of scalability and generalization for end-to-end automatic speech recognition”, by applied scientist Egor Lakomkin and his Amazon colleagues, investigates the regularization of subwords in the model, or the enforcement of greater consistency in how words are segmented into subwords. In experiments, the researchers show that using multiple segmentations of the same speech transcription during training can reduce the ASR error rate by 8.4% in a model trained on 5,000 hours of speech data.

Efficient minimum word error rate training of RNN-transducer for end-to-end speech recognition”, by applied scientist Jinxi Guo and six of his Amazon colleagues, investigates a novel loss function — an evaluation criterion during training — for such RNN-T ASR systems. In experiments, it reduced the systems’ error rates by 3.6% to 9.2%.

For each input, RNN-Ts output multiple possible solutions — or hypotheses — ranked according to probability. In ASR applications, RNN-Ts are typically trained to maximize the probabilities they assign the correct transcriptions of the input speech.

But trained speech recognizers are judged, by contrast, according to their word error rates, or the rate at which they make mistakes — misinterpretations, omissions, or erroneous insertions. Jinxi Guo and his colleagues investigated efficient ways to directly train an RNN-T ASR system to minimize word error rate.

That means, for each training example, minimizing the expected word errors of the most probable hypotheses. But computing the probabilities of those hypotheses isn’t as straightforward as it may sound.

That’s because the exact same sequence of output subwords can align with the sequence of input frames in different ways: one output sequence, for instance, might identify the same subword as having begun one frame earlier or later than another output sequence does. Computing the probability of a hypothesis requires summing the probabilities of all its alignments.

The brute-force solution to this problem would be computationally impractical. But Guo and his colleagues propose using the forward-backward algorithm, which exploits the overlaps between alignments, storing intermediate computations that can be re-used. The result is a computationally efficient algorithm that enables a 3.6% to 9.2% reduction in error rates for various RNN-T models.

The other Amazon ASR papers at this year’s Interspeech are

DiPCo - Dinner Party Corpus
Maarten Van Segbroeck, Zaid Ahmed, Ksenia Kutsenko, Cirenia Huerta, Tinh Nguyen, Björn Hoffmeister, Jan Trmal, Maurizio Omologo, Roland Maas

End-to-end neural transformer based spoken language understanding
Martin Radfar, Athanasios Mouchtaris, Siegfried Kunzmann

Improving speech recognition of compound-rich languages
Prabhat Pandey, Volker Leutnant, Simon Wiesler, Jahn Heymann, Daniel Willett

Improved training strategies for end-to-end speech recognition in digital voice assistants
Hitesh Tulsiani, Ashtosh Sapru, Harish Arsikere, Surabhi Punjabi, Sri Garimella

Leveraging unlabeled speech for sequence discriminative training of acoustic models
Ashtosh Sapru, Sri Garimella

Quantization aware training with absolute-cosine regularization for automatic speech recognition
Hieu Duy Nguyen, Anastasios Alexandridis, Athanasios Mouchtaris

Rescore in a flash: Compact, cache efficient hashing data structures for N-gram language models
Grant P. Strimel, Ariya Rastrow, Gautam Tiwari, Adrien Pierard, Jon Webb

Semantic complexity in end-to-end spoken language understanding
Joseph McKenna, Samridhi Choudhary, Michael Saxon, Grant P. Strimel, Athanasios Mouchtaris

Speech to semantics: Improve ASR and NLU jointly via all-neural interfaces
Milind Rao, Anirudh Raju, Pranav Dheram, Bach Bui, Ariya Rastrow 

About the Author
Björn Hoffmeister is the director of applied science for Alexa Speech.

Related content

View from space of a connected network around planet Earth representing the Internet of Things.
Get more from Amazon Science
Sign up for our monthly newsletter

Work with us

See more jobs
US, MA, Cambridge
Alexa is the groundbreaking voice service that powers Echo and other devices designed around your voice. Our team is creating the science and technology behind Alexa. We’re working hard, having fun, and making history. Come join our team! You will have an enormous opportunity to impact the customer experience, design, architecture, and implementation of a cutting edge product used every day by people you know.We’re looking for a passionate, talented, and inventive scientist to help build industry-leading conversational technologies that customers love. Our team's mission is the enable Alexa to understand sounds and vocalization beyond speech. As a Research Scientist, you will work with talented peers to develop novel algorithms and modeling techniques to advance the state of the art in speech and audio processing. Your work will directly impact our customers in the form of novel products and services that make use of speech and language technology. You will leverage Amazon’s heterogeneous data sources and large-scale computing resources to accelerate advances in spoken language understanding. You will mentor junior scientists, create and drive new initiatives.
DE, Aachen
Amazon Alexa is looking for a Senior Applied Science Manager to lead the development of next generation spoken and visual understanding systems that will revolutionize the revolutionize the way developers build skills and enable customers to achieve complex and high value goals in their everyday use of Alexa. In this role, you will manage teams of passionate, talented, and inventive scientists, to develop industry-leading Computer Vision (CV), Natural Language Processing (NLP), and Dialog Management, and drive them successfully to production for the benefit of millions of Alexa users. Your mission is to push the envelope for CV, NLP and Dialog Management systems in order to provide the industry-leading, best-possible experience for our customers.As a Senior Applied Science Manager, you will identify research directions, create roadmaps for forward-looking research and communicate them to senior leadership, and work closely with engineering teams to bring research to production. You will work with teams of talented scientists, and fill the ranks by attracting the best scientists in CV, NLP, and Dialog areas, by representing Amazon Alexa at international science conferences. You will work with talented peers and leverage Amazon’s heterogeneous data sources and large-scale computing resources.
US, NY, New York
Are you interested in big data, machine learning, and product recommendations? If so, the Product Semantics team in Amazon Product Graph might be the right place for you. We are a team in a fast-paced organization with a huge impact on hundreds of millions of customers. We innovate at the intersection of customer experience, deep learning, and high-scale machine-learning systems.As the world’s most customer-centric company, Amazon heavily invests in inventing and applying state-of-art technologies to build world-class product recommendation systems to improve shopper experience. We break fresh ground to create world-class customer-facing features to help customers discover high quality products that meet their needs, and provide most relevant product information to help customers make confident shopping decisions. We are a highly motivated, collaborative, and fun-loving team with a strong entrepreneurial spirit and bias for action. With a broad mandate to experiment and innovate, we have a very wide range of new opportunities to explore.The Product Semantics team in Amazon Product Graph, based in Seattle and New York City, is looking for scientists who love big data, are passionate about understanding products and product relationships from product profiles, reviews, and search log, and who are capable of inventing and applying Machine Learning, NLP, and Computer Vision techniques that will leave no valuable data behind. Our applied scientists work closely with software engineers to put algorithms into practice. They also work in partnership with teams across Amazon to create enormous benefits for our customers.The Product Semantics team in Amazon Product Graph, based in Seattle and New York City, is looking for Applied Scientists who love big data, are passionate about understanding products and product relationships from product profiles, reviews, and search log, and who are capable of inventing and applying Machine Learning, NLP, and Computer Vision techniques to solve real world problem and serve customer-facing features on Amazon. Our applied scientists work closely with software engineers to put algorithms into practice. They also work in partnership with teams across Amazon to create enormous benefits for our customers.If you have an entrepreneurial spirit, know how to deliver, love to work with data, are deeply technical, highly innovative and long for the opportunity to build solutions to challenging problems that directly impact the company's bottom-line, we want to talk to you.Key Responsibilities- Use machine learning and analytical techniques to create scalable solutions for business problems- Analyze and extract relevant information from large amounts of Amazon's historical business data to help automate and optimize key processes- Design, development and evaluation of highly innovative models for predictive learning- Work closely with software engineering teams to drive model implementations and new feature creations- Establish scalable, efficient, automated processes for large scale data analyses, model development, model validation and model implementation- Research and implement novel machine learning and statistical approaches
US, NY, New York
Are you interested in big data, machine learning, and product recommendations? If so, the Product Semantics team in Amazon Product Graph might be the right place for you. We are a team in a fast-paced organization with a huge impact on hundreds of millions of customers. We innovate at the intersection of customer experience, deep learning, and high-scale machine-learning systems.As the world’s most customer-centric company, Amazon heavily invests in inventing and applying state-of-art technologies to build world-class product recommendation systems to improve shopper experience. We break fresh ground to create world-class customer-facing features to help customers discover high quality products that meet their needs, and provide most relevant product information to help customers make confident shopping decisions. We are a highly motivated, collaborative, and fun-loving team with a strong entrepreneurial spirit and bias for action. With a broad mandate to experiment and innovate, we have a very wide range of new opportunities to explore.The Product Semantics team in Amazon Product Graph, based in Seattle and New York City, is looking for scientists who love big data, are passionate about understanding products and product relationships from product profiles, reviews, and search log, and who are capable of inventing and applying Machine Learning, NLP, and Computer Vision techniques that will leave no valuable data behind. Our applied scientists work closely with software engineers to put algorithms into practice. They also work in partnership with teams across Amazon to create enormous benefits for our customers.The Product Semantics team in Amazon Product Graph, based in Seattle and New York City, is looking for Applied Scientists who love big data, are passionate about understanding products and product relationships from product profiles, reviews, and search log, and who are capable of inventing and applying Machine Learning, NLP, and Computer Vision techniques to solve real world problem and serve customer-facing features on Amazon. Our applied scientists work closely with software engineers to put algorithms into practice. They also work in partnership with teams across Amazon to create enormous benefits for our customers.If you have an entrepreneurial spirit, know how to deliver, love to work with data, are deeply technical, highly innovative and long for the opportunity to build solutions to challenging problems that directly impact the company's bottom-line, we want to talk to you.Key Responsibilities- Use machine learning and analytical techniques to create scalable solutions for business problems- Analyze and extract relevant information from large amounts of Amazon's historical business data to help automate and optimize key processes- Design, development and evaluation of highly innovative models for predictive learning- Work closely with software engineering teams to drive model implementations and new feature creations- Establish scalable, efficient, automated processes for large scale data analyses, model development, model validation and model implementation- Research and implement novel machine learning and statistical approaches
US, WA, Seattle
We are a team of doers working passionately to apply cutting-edge advances in technology to solve real-world problems. As a Research Scientist, you will work with a unique and gifted team developing exciting products for consumers and collaborate with cross-functional teams. Our team rewards intellectual curiosity while maintaining a laser-focus in bringing products to market. Competitive candidates are responsive, flexible, and able to succeed within an open, collaborative, entrepreneurial, startup-like environment. At the cutting edge of both academic and applied research in this product area, you have the opportunity to work together with some of the most talented scientists, engineers, and product managers.Inclusive Team Culture:Here at Amazon, we embrace our differences. We are committed to furthering our culture of inclusion. We have ten employee-led affinity groups, reaching 40,000 employees in over 190 chapters globally. We have innovative benefit offerings, and host annual and ongoing learning experiences, including our Conversations on Race and Ethnicity (CORE) and AmazeCon (gender diversity) conferences. Amazon’s culture of inclusion is reinforced within our 14 Leadership Principles, which remind team members to seek diverse perspectives, learn and be curious, and earn trust.Work/Life BalanceOur team puts a high value on work-life balance. It isn’t about how many hours you spend at home or at work; it’s about the you establish that brings energy to both parts of your life. We believe striking the right balance between your personal and professional life is critical to life-long happiness and fulfillment. We offer flexibility in working hours and encourage you to find your own balance between your work and personal lives.Mentorship & Career GrowthOur team is dedicated to supporting new members. We have a broad mix of experience levels and tenures, and we’re building an environment that celebrates knowledge sharing and mentorship. Our senior members enjoy one-on-one mentoring and thorough, but kind, code reviews. We care about your career growth and strive to assign projects based on what will help each team member develop into a better-rounded engineer and enable them to take on more complex tasks in the future.
GB, MLN, Edinburgh
Do you want to make a real difference to real people's lives? Want to design and build fair and explainable systems which automate recruitment processes across Amazon? Come and be part of a team that develops new machine learning (ML) technologies, which help Amazon scale for its customers by recruiting diverse teams.A day in the lifeAmazon is looking for an Applied Scientist. As an scientist working at Amazon, you will play a key role in identifying business opportunities, measuring opportunity, inventing and prototyping solutions. You will use a wide range of technologies, programming languages and systems, and work alongside an experienced engineering team. You will have the freedom and encouragement to explore your own ideas and the reward of seeing your contributions benefit millions of customers worldwide.About the hiring groupJoin a team full of talented people who come from all over the world. Enjoy the chance to work in a relaxed setting with a good social life. The team, primarily based in Edinburgh, Scotland, is rapidly expanding.Job responsibilitiesYou will work as an ML scientist in a team of other scientists and software developers. You will primarily be writing solutions in Python and will be using the latest technologies including AWS (e.g. Sagemaker). You will be contributing regularly to the code base as this is an applied role with the expectation of 50% of your time spent on the code. Your solutions will meet remarkably high standards of performance and reliability, and will operate at massive scale.You will work as part of a sustainably paced agile team. You will play a hands on role in your team giving you the responsibility, authority, and autonomy to ensure success. You will be involved in every aspect of the process - from idea generation, customer engagement, business analysis and scientific design through to software development and operations.We are looking for ML scientists who can delight our customers by continually learning and inventing and who want to use their expertise to solve real business problems. Our ideal candidate has a strong scientific foundation and experience of statistical analysis and model building and has a passion for fairness and explainability in ML systems.The role offers an exceptional opportunity for growth and to make a real difference to Amazon recruitment. If you are selected, you have the opportunity to really impact our business by inventing, improving, and building world class systems, delivering results, working on exciting and challenging projects.
US, WA, Bellevue
Do you enjoy solving complex problems? Are you eager to change the world with data science? At Amazon Taskless, we challenge ourselves with questions like, what if we can verify documentation in seconds instead of days? What if we could quickly automate complex processes which are not well documented? What if we can improve customer retention?By adopting technologies such as machine learning, computer vision (Amazon Rekognition & Textract) and natural language processing(Amazon Lex), Amazon Taskless transforms tedious businesses processing with Intelligent Automation and Robotic Process Automation. We built an identity management system, which simplify compliance across all Amazon businesses including Twitch, Flex, Amazon sellers, Kindle Direct Publishing authors globally.As a Data Scientist, you will work on our Science team and partner closely with other data scientists , data engineers as well as product managers, UX designers, and business partners across Amazon to accurately model and remove tasks from their processes. Outputs from your models will directly improve customer experience across Amazon while delivering cost savings. You will be responsible for building data science prototypes that optimize business processes and innovate for our customers in new ways.You are skeptical. When someone gives you a data source or walks you through their process, you pepper them with questions about, accuracy, coverage, and the need of steps in their process. When you’re told a model can make assumptions, you aggressively try to break those assumptions.You do whatever it takes to add value. You don’t care whether you’re building complex machine learning models, writing blazing fast code, integrating multiple disparate data-sets, or creating baseline models - you care passionately about stakeholders and know that as a curator of data insight you can unlock massive cost savings and retain customers.You have a limitless curiosity. You constantly ask questions about the technologies and approaches we are taking and are constantly learning about industry best practices you can bring to our team.You have excellent business and communication skills to be able to work with product owners to understand key business questions and earn the trust of senior leaders. You will need to make the complex simple to understand.You are comfortable juggling competing priorities and handling ambiguity. You thrive in an agile and fast-paced environment on highly visible projects and initiatives. The tradeoffs of cost savings and customer experience are constantly up for debate among senior leadership - you will help drive this conversation.
US, WA, Bellevue
Do you enjoy solving complex problems? Are you eager to change the world with data science? At Amazon Taskless, we challenge ourselves with questions like, what if we can verify documentation in seconds instead of days? What if we could quickly automate complex processes which are not well documented? What if we can improve customer retention?By adopting technologies such as machine learning, computer vision (Amazon Rekognition & Textract) and natural language processing(Amazon Lex), Amazon Taskless transforms tedious businesses processing with Intelligent Automation and Robotic Process Automation. We built an identity management system, which simplify compliance across all Amazon businesses including Twitch, Flex, Amazon sellers, Kindle Direct Publishing authors globally.As a Data Scientist, you will work on our Science team and partner closely with other data scientists , data engineers as well as product managers, UX designers, and business partners across Amazon to accurately model and remove tasks from their processes. Outputs from your models will directly improve customer experience across Amazon while delivering cost savings. You will be responsible for building data science prototypes that optimize business processes and innovate for our customers in new ways.You are skeptical. When someone gives you a data source or walks you through their process, you pepper them with questions about, accuracy, coverage, and the need of steps in their process. When you’re told a model can make assumptions, you aggressively try to break those assumptions.You do whatever it takes to add value. You don’t care whether you’re building complex machine learning models, writing blazing fast code, integrating multiple disparate data-sets, or creating baseline models - you care passionately about stakeholders and know that as a curator of data insight you can unlock massive cost savings and retain customers.You have a limitless curiosity. You constantly ask questions about the technologies and approaches we are taking and are constantly learning about industry best practices you can bring to our team.You have excellent business and communication skills to be able to work with product owners to understand key business questions and earn the trust of senior leaders. You will need to make the complex simple to understand.You are comfortable juggling competing priorities and handling ambiguity. You thrive in an agile and fast-paced environment on highly visible projects and initiatives. The tradeoffs of cost savings and customer experience are constantly up for debate among senior leadership - you will help drive this conversation.
US, WA, Bellevue
Do you enjoy solving complex problems? Are you eager to change the world with data science? At Amazon Taskless, we challenge ourselves with questions like, what if we can verify documentation in seconds instead of days? What if we could quickly automate complex processes which are not well documented? What if we can improve customer retention?By adopting technologies such as machine learning, computer vision (Amazon Rekognition & Textract) and natural language processing(Amazon Lex), Amazon Taskless transforms tedious businesses processing with Intelligent Automation and Robotic Process Automation. We built an identity management system, which simplify compliance across all Amazon businesses including Twitch, Flex, Amazon sellers, Kindle Direct Publishing authors globally.As a Data Scientist, you will work on our Science team and partner closely with other data scientists , data engineers as well as product managers, UX designers, and business partners across Amazon to accurately model and remove tasks from their processes. Outputs from your models will directly improve customer experience across Amazon while delivering cost savings. You will be responsible for building data science prototypes that optimize business processes and innovate for our customers in new ways.You are skeptical. When someone gives you a data source or walks you through their process, you pepper them with questions about, accuracy, coverage, and the need of steps in their process. When you’re told a model can make assumptions, you aggressively try to break those assumptions.You do whatever it takes to add value. You don’t care whether you’re building complex machine learning models, writing blazing fast code, integrating multiple disparate data-sets, or creating baseline models - you care passionately about stakeholders and know that as a curator of data insight you can unlock massive cost savings and retain customers.You have a limitless curiosity. You constantly ask questions about the technologies and approaches we are taking and are constantly learning about industry best practices you can bring to our team.You have excellent business and communication skills to be able to work with product owners to understand key business questions and earn the trust of senior leaders. You will need to make the complex simple to understand.You are comfortable juggling competing priorities and handling ambiguity. You thrive in an agile and fast-paced environment on highly visible projects and initiatives. The tradeoffs of cost savings and customer experience are constantly up for debate among senior leadership - you will help drive this conversation.
US, WA, Bellevue
Do you enjoy solving complex problems? Are you eager to change the world with data science? At Amazon Taskless, we challenge ourselves with questions like, what if we can verify documentation in seconds instead of days? What if we could quickly automate complex processes which are not well documented? What if we can improve customer retention?By adopting technologies such as machine learning, computer vision (Amazon Rekognition & Textract) and natural language processing(Amazon Lex), Amazon Taskless transforms tedious businesses processing with Intelligent Automation and Robotic Process Automation. We built an identity management system, which simplify compliance across all Amazon businesses including Twitch, Flex, Amazon sellers, Kindle Direct Publishing authors globally.As a Data Scientist, you will work on our Science team and partner closely with other data scientists , data engineers as well as product managers, UX designers, and business partners across Amazon to accurately model and remove tasks from their processes. Outputs from your models will directly improve customer experience across Amazon while delivering cost savings. You will be responsible for building data science prototypes that optimize business processes and innovate for our customers in new ways.You are skeptical. When someone gives you a data source or walks you through their process, you pepper them with questions about, accuracy, coverage, and the need of steps in their process. When you’re told a model can make assumptions, you aggressively try to break those assumptions.You do whatever it takes to add value. You don’t care whether you’re building complex machine learning models, writing blazing fast code, integrating multiple disparate data-sets, or creating baseline models - you care passionately about stakeholders and know that as a curator of data insight you can unlock massive cost savings and retain customers.You have a limitless curiosity. You constantly ask questions about the technologies and approaches we are taking and are constantly learning about industry best practices you can bring to our team.You have excellent business and communication skills to be able to work with product owners to understand key business questions and earn the trust of senior leaders. You will need to make the complex simple to understand.You are comfortable juggling competing priorities and handling ambiguity. You thrive in an agile and fast-paced environment on highly visible projects and initiatives. The tradeoffs of cost savings and customer experience are constantly up for debate among senior leadership - you will help drive this conversation.
US, WA, Bellevue
At Amazon we're working to be the most customer-centric company on earth. Within the Access Points team, we do this by creating delivery experiences that delight customers, growing our worldwide network of Amazon Hub Lockers and Counters, providing away from home pickup options, and by creating new delivery initiatives that solve the changing needs of our Customers. At any Access Point, customers should expect to return, or redirect their Amazon deliveries. We measure our impact in transportation savings, revenue, and downstream customer acquisition / engagement /purchasing with Amazon.We are looking for an accomplished Manager, Research Science for Amazon Access Point’s worldwide data science team. You will define the research science direction for the team and work with our engineers to create an advanced system solving mathematically complex constraint problems. You will lead the team to own development of novel algorithmic architectures, toward the ultimate goal of accurately predicting customer purchase propensity, demand pattern and optimizing for site selection topology future Access Point locations and eligible products worldwide.Access Point has a rapidly growing customer base and an exciting science charter in front of us that includes solving highly complex algorithmic problems. You will work closely with and learn from data professionals from various disciplines (eg data engineers, analysts, machine learning engineers, economists and other fellow research scientists).Key responsibilities:· Hire, manager and grow a team of scientists and be the thought leader on the team· Collaborate with product managers and engineering teams to design and implement software solutions for Amazon problems· Contribute to progress of the Amazon and broader research communities by producing publications· Be hands-on when needed, to mine the large amount of data, prototype and implement new learning algorithms and prediction techniques to improve forecast accuracy or optimization performance
US, PA, Pittsburgh
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 push the envelope in Automatic Speech Recognition (ASR), Natural Language Understanding (NLU), and Audio Signal Processing, in order to provide the best-possible experience for our customers.As a Scientist, you will work with talented peers to develop novel algorithms and modeling techniques to advance the state of the art in spoken language understanding. Your work will directly impact our customers in the form of products and services that make use of speech and language technology. You will leverage Amazon’s heterogeneous data sources and large-scale computing resources to accelerate advances in spoken language understanding.We are hiring in all areas of spoken language understanding: ASR, NLU, text-to-speech (TTS), and Dialog Management.Position Responsibilities:· Participate in the design, development, evaluation, deployment and updating of data-driven models and analytical solutions for machine learning (ML) and/or natural language (NL) applications.· Develop and/or apply statistical modeling techniques (e.g. Bayesian models and deep neural networks), optimization methods, and other ML techniques to different applications in business and engineering.· Routinely build and deploy ML models on available data.· Research and implement novel ML and statistical approaches to add value to the business. Mentor junior engineers and scientists.
CA, ON, Toronto
Amazon Sponsored Ads is one of the fastest growing business domains and we are looking for talented scientists to join this team of incredible scientists to contribute to this growth. 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 team of machine learning and data scientists, with an entrepreneurial spirit and bias for action. We have a broad mandate to experiment and innovate, and we are growing at an unprecedented rate with a seemingly endless range of new opportunities. Sponsored Products (SP) Bids and Budgets team is focussed on helping advertisers set their campaign bids and budgets in an optimized fashion.As an Research Scientist on this team you will:· Build machine learning models and utilize data analysis to deliver scalable solutions to business problems.· Perform hands-on analysis and modeling with very large data sets to develop insights that increase traffic monetization and merchandise sales without compromising shopper experience.· Work closely with software engineers on detailed requirements, technical designs and implementation of end-to-end solutions in production.· Design and run A/B experiments that affect hundreds of millions of customers, evaluate the impact of your optimizations and communicate your results to various business stakeholders.· Work with scientists and economists to model the interaction between organic sales and sponsored content and to further evolve Amazon's marketplace.· Establish scalable, efficient, automated processes for large-scale data analysis, machine-learning model development, model validation and serving.· Research new predictive learning approaches for the sponsored products business.Why you love this opportunityAmazon is investing heavily in building a world-class advertising business. This team is responsible for defining and delivering a collection of advertising products that drive discovery and sales. Our solutions generate billions in revenue and drive long-term growth for Amazon’s Retail and Marketplace businesses. We deliver billions of ad impressions, millions of clicks daily, and break fresh ground to create world-class products. We are highly motivated, collaborative, and fun-loving team with an entrepreneurial spirit - with a broad mandate to experiment and innovate.Impact and Career GrowthYou will invent new experiences and influence customer-facing shopping experiences to help suppliers grow their retail business and the auction dynamics that leverage native advertising; this is your opportunity to work within the fastest-growing businesses across all of Amazon! Define a long-term science vision for our advertising business, driven fundamentally from our customers' needs, translating that direction into specific plans for research and applied scientists, as well as engineering and product teams. This role combines science leadership, organizational ability, technical strength, product focus, and business understanding.Team video https://youtu.be/zD_6Lzw8raE
US, MA, Cambridge
Alexa is Amazon’s intelligent cloud-based voice recognition and natural language understanding virtual assistant. We’re building the speech and language solutions behind Amazon Alexa and other Amazon products and services. Come join our team and help improve the customer experience for the growing base of Alexa users!The Alexa Artificial Intelligence (AI) team is seeking a talented Applied Scientist to build ML models to detect issues that end-users have in their interactions with Alexa (defects and their possible root causes). These models are then used to monitor trends over time with Customer Experience (CX) metrics, guardrail metrics in weblabs, setting defect reduction goals, and defect discovery and resolution.A day in the life· Design, build, test and release predictive ML models· Ensure data quality throughout all stages of acquisition and processing, including such areas as data sourcing/collection, ground truth generation, normalization, and transformation.· Collaborate with colleagues from science, engineering and business backgrounds.· Present proposals and results to partner teams in a clear manner backed by data and coupled with actionable conclusions· Work with engineers to develop efficient data querying and inference infrastructure for both offline and online use casesAbout the hiring groupAlexa AI is an analytics and science team within Alexa. Our mission is to provide an understanding of the customer experience that allows Alexa teams to improve system performance and customer engagement. Our primary deliverables are CX metrics, analytics tools, and customer insights.Job responsibilitiesAs an Applied Scientist with our Alexa AI team, you will work on assessing Alexa's performance using predictive ML models. You will build and improve models to classify Alexa’s responses as correct/incorrect, and predict the most likely cause of failure in cases of incorrect action. Your work will directly impact our customers in the form of products and services that make use of speech and language technology, particularly in developing predictive models to continuously improve the Alexa experience for our customers.Amazon is committed to a diverse and inclusive workplace. Amazon is an equal opportunity employer and does not discriminate on the basis of race, national origin, gender, gender identity, sexual orientation, protected veteran status, disability, age, or other legally protected status. For individuals with disabilities who would like to request an accommodation, please visit https://www.amazon.jobs/en/disability/us.
LU, Luxembourg
Are you a talented and inventive engineer with strong passion about Artificial Intelligence and Predictive Modeling? Would you like to develop Machine-Learning tools by playing a key role within EU RME Predictive Analytics team? Our mission is to drive the Predictive Maintenance (PdM) and Spare Parts (SP) programs for Amazon EU Operations that consists of complex automation, sortation, robotic and materials handling systems.As Machine Learning Tool Specialist you will be working with large distributed systems of data and providing predictive maintenance expertise for over 2000 maintenance engineers, managers and administrators by supporting the entire network managed by EU RME, which may include non-EU locations (such as Singapore, Australia and Japan). You will connect with world leaders in your field and you will be tackling ML challenges by carrying out a systematic review of existing solutions. The appropriate choice of the ML methods and their deployment into effective tools will be the key for the success in this role.The successful candidate will be a self-starter comfortable with ambiguity, with strong attention to detail and outstanding ability in balancing technical leadership with strong business judgment to make the right decisions about model and method choices.Key Areas of Responsibilities:· Provide technical expertise to support team strategies that will take EU RME towards World Class predictive maintenance practices and processes, driving better equipment up-time and lower repair costs with optimized spare parts inventory and placement· Implement an advanced maintenance framework utilizing Machine Learning technologies to drive equipment performance leading to reduced unplanned downtime· Provide technical expertise to support the development of long-term spares management strategies that will ensure spares availability at an optimal level for local sites and reduce the cost of spares
LU, Luxembourg
Are you a talented and inventive engineer with strong passion about Artificial Intelligence and Predictive Modeling? Would you like to develop Machine-Learning tools by playing a key role within EU RME Predictive Analytics team? Our mission is to drive the Predictive Maintenance (PdM) and Spare Parts (SP) programs for Amazon EU Operations that consists of complex automation, sortation, robotic and materials handling systems.As Machine Learning Tool Specialist you will be working with large distributed systems of data and providing predictive maintenance expertise for over 2000 maintenance engineers, managers and administrators by supporting the entire network managed by EU RME, which may include non-EU locations (such as Singapore, Australia and Japan). You will connect with world leaders in your field and you will be tackling ML challenges by carrying out a systematic review of existing solutions. The appropriate choice of the ML methods and their deployment into effective tools will be the key for the success in this role.The successful candidate will be a self-starter comfortable with ambiguity, with strong attention to detail and outstanding ability in balancing technical leadership with strong business judgment to make the right decisions about model and method choices.Key Areas of Responsibilities:· Provide technical expertise to support team strategies that will take EU RME towards World Class predictive maintenance practices and processes, driving better equipment up-time and lower repair costs with optimized spare parts inventory and placement· Implement an advanced maintenance framework utilizing Machine Learning technologies to drive equipment performance leading to reduced unplanned downtime· Provide technical expertise to support the development of long-term spares management strategies that will ensure spares availability at an optimal level for local sites and reduce the cost of spares
US, WA, Seattle
Do you want to join Alexa Artificial Intelligence (AI), the science team behind Amazon’s intelligence voice assistance system? Do you want to utilize cutting-edge deep-learning and machine learning algorithms to delight millions of Alexa users around the world?If your answers to these questions are “yes”, then come join the Alexa AI team, which is in charge of improving Alexa user satisfaction through real-time metrics monitoring and continuous closed-loop learning. The team owns the modules that reduce user perceived defects and frictions through utterance reformulation, contextual and personalized hypothesis ranking.A day in the lifeAs a Senior Applied Scientist, you will be working alongside a team of experienced machine/deep learning scientists and engineers to create data driven machine learning models and solutions on tasks such as sequence-to-sequence query reformulation, graph feature embedding, personalized ranking, etc..About the hiring groupThe Alexa AI team is in charge of improving Alexa user satisfaction through real-time metrics monitoring and continuous closed-loop learning. The team owns the modules that reduce user perceived defects and frictions through utterance reformulation, contextual and personalized hypothesis ranking.Job responsibilitiesYou will be expected to:· Analyze, understand, and model user-behavior and the user-experience based on large scale data, to detect key factors causing satisfaction and dissatisfaction (SAT/DSAT).· Build and measure novel online & offline metrics for personal digital assistants and user scenarios, on diverse devices and endpoints· Create and innovate deep learning and/or machine learning based algorithms for utterance reformulation and contextual hypothesis ranking to reduce user dissatisfaction in various scenarios;· Perform model/data analysis and monitor user-experienced based metrics through online A/B testing;· Research and implement novel machine learning and deep learning algorithms and models.Amazon is committed to a diverse and inclusive workplace. Amazon is an equal opportunity employer and does not discriminate on the basis of race, national origin, gender, gender identity, sexual orientation, protected veteran status, disability, age, or other legally protected status. For individuals with disabilities who would like to request an accommodation, please visit https://www.amazon.jobs/en/disability/us.
US, WA, Seattle
The Fresh Food Fast organization is responsible for transforming the online and offline grocery experience for Amazon. We are seeking a senior science leader to define our long-term science vision, build out a high-performing team and deliver business critical scientific models to increase customer engagement, inform long-term investment decisions, and measure how grocery is contributing to Prime and Amazon.A day in the life· You will influence senior leaders (VP+) across business, product, finance, and engineering functions and you will partner closely with central Amazon teams to pioneer new models to measure grocery’s future impact to Prime and Amazon.· You will manage a team of Data Scientists, Economists and BIEs to deliver results on behalf of customersAbout the hiring groupWe’re a team of Product Managers, Data Scientists, Economists and Business Intelligence Engineers focused on deeply understanding how F3 customers engage with physical and online grocery stores in order to enhance their shopping experience, drive engagement and loyalty, and measure their long-term impact to Amazon.Job responsibilitiesYour team will apply complex scientific methods to challenging business problems including, “How can we encourage customers to shop more frequently?”, and “how should we measure the impact of physical store expansion and technology innovation in those stores (e.g. Just Walk Out Technology)?”. You will power through ambiguity, finding the right solutions to problems and influencing others to align with your approach and help drive results. You will mentor and develop scientists to achieve their goals, raising the bar technically and driving scale and efficiencies to better leverage our data and technologies.Amazon is committed to a diverse and inclusive workplace. Amazon is an equal opportunity employer and does not discriminate on the basis of race, national origin, gender, gender identity, sexual orientation, protected veteran status, disability, age, or other legally protected status. For individuals with disabilities who would like to request an accommodation, please visit https://www.amazon.jobs/en/disability/us.
DE, BE, Berlin
Our mission is to build the automated intelligence supporting critical service operations at global scale. The Intelligent Cloud Control Machine Learning (ICCML) team works to automate complex large-scale operations of Amazon’s consumer services by developing data-driven, scalable, and seamless solutions available to customers and ICC partners. We employ machine learning to reduce system and information complexity while improving service reliability. We invent practical approaches within application areas such as anomaly detection, time series analysis, classification, causal inference, and text mining, and we apply the latest and most sound techniques of probabilistic modelling, estimation, deep neural networks, and natural language processing (NLP). Working with us offers exciting challenges where you will grow as an applied scientist and technical leader, combining your scientific and engineering skills to solve complex machine learning problems together with our tech teams around the world.As an Applied Scientist of the ICCML team, you will have the important role of mapping business problems to high-impact solutions. You will turn theoretically sound methods into practically applicable models designed for processing massive volumes of data in large-scale environments. You will define business relevant solutions implemented as end-to-end machine learning functions and data processing pipelines that integrate with our partners production systems. In a fast-paced innovation environment, you will work closely with our Applied Scientists, Machine Learning Engineers, and partners to design machine learning models and experiments at scale. You dive deep into all aspects of the practical machine learning development cycle, encompassing sound use of data pre-processing techniques, analysis, modelling, and validation methods. You master the complex theory under the hood of machine learning and you keep up to date with the latest scientific development in information processing, modelling, and learning methods. You take lead of the scientific and technical work in cross-team collaborations with the ultimate objective of creating a delightful experience for our customers using our services.
IL, Tel Aviv
You: Alexa, I am looking for a new career opportunity, where I could conduct applied research, impact millions of customers, and publish about it in top conferences. What do you suggest?Alexa: The Alexa Shopping team is looking for brilliant applied researchers to help me become the best personal shopping assistant. Do you want to hear more?You: Yes, please!Alexa: As an applied researcher in the Alexa Shopping Research team, you will be responsible for research, design, and implementation of new AI technologies for voice assistants. You will collaborate with scientists, engineers, and product partners locally and abroad. Your work will inventing, experimenting with, and launching new features, products and systems. Ideally you have a expertise in at least one of the following fields: Web search & data mining, Machine Learning, Natural Language Processing, Computer Vision, Speech Processing or Artificial Intelligence, with both hands-on experience and publications at top relevant academic venues.