Low-precision arithmetic makes robot localization more efficient

Using different levels of precision for different arithmetic tasks reduces computational burden without compromising performance.

Simultaneous localization and mapping (SLAM) is the core technology of autonomous mobile robots. It involves simultaneously building a map of the robot’s environment and finding the robot’s location within that map.

SLAM is computationally intensive, and deploying it on resource-constrained robots — such as consumer household robots — generally requires techniques for making computations more tractable.

Related content
Two Alexa AI papers present novel methodologies that use vision and language understanding to improve embodied task completion in simulated environments.

One such technique is the use of low-precision floating-point arithmetic, or reducing the number of bits used to represent numbers with decimal points. The technique is popular in deep learning, where halving the number of bits (from the standard 32 to 16) can double computational efficiency with little effect on accuracy.

But applying low-precision arithmetic to SLAM is more complicated. Where deep-learning-based classification models are discrete-valued, SLAM involves solving a nonlinear optimization problem with continuous-valued functions, which require higher accuracy.

At Amazon, we’ve tackled this problem by designing a novel mixed-precision solver, which combines 64-bit (fp64), 32-bit (fp32), and 16-bit (fp16) precisions for nonlinear optimization problems in the SLAM algorithm. This innovation paves the way for faster and greener on-device navigation.

General framework

A SLAM algorithm has two key components: visual odometry and loop closure. Visual odometry gives real-time estimates of the robot’s pose, or its orientation and location on the map, based on the most recent observations. When the robot recognizes that it has arrived at a place that it previously visited, it closes the loop by globally correcting its map and its location estimate.

Related content
A model that estimates depth from 2-D images learns to adjust to differences between images produced by different cameras, reducing error by about 20%.

Both visual odometry and loop closure involve solving nonlinear optimization problems — bundle adjustment (BA) and pose graph optimization (PGO), respectively. To solve them efficiently, SLAM systems typically use approximate methods that recast them as sequences of linearized optimization problems. If the goal is to find the pose estimate x, then each linear problem minimizes the linearized error function, which is the sum of the current error function and its first-order correction. The first-order correction is the product of the Jacobian, which is the matrix of the function’s first-order derivatives, and the update to the pose estimation. The linear problems are typically solved through factorization, using either Cholesky or QR methods. The solution of each linearized optimization problem is the update for the current pose estimate.

The general procedure is to start with the current approximation of x, compute the error function and the Jacobian, solve a linear optimization problem, and update x accordingly, repeating the process until certain stopping criteria are met. At each iteration, the value of the error function is known as the residual, since it’s the residual error left over from the previous iteration.

General framework.png
General framework for mixed-precision nonlinear optimization.

The most expensive computations in the nonlinear optimizations for both BA and PGO are the computation of the Jacobian (about 15% of the optimization time) and the solution of the linear problem (about 60%). Simply solving either problem at half-precision (fp16) from beginning to end will result in lower accuracy and sometimes numerical instability.

To mitigate these difficulties, we regularize and scale the matrices to avoid overflow and rank deficiency. The rank deficiency occurs when columns of the Jacobian are linearly dependent. Through careful experiments, we further identified the computations to be done at precision higher than fp16 and proposed a mixed-precision nonlinear optimization solver.

Related content
Deep learning to produce invariant representations, estimations of sensor reliability, and efficient map representations all contribute to Astro’s superior spatial intelligence.

We found that, to match the accuracy of the solution in pure double-precision, the following two components have to be computed in precision higher than fp16:

  • The residual must be evaluated in single or higher precision;
  • The update of x, which is a six-degree position-angle update, must be done in double precision.

Although this general optimization framework applies to both BA and PGO, the details vary across the two applications, because of the different structures and properties of the matrices in the linear problems. We thus propose two mixed-precision solving strategies for the relevant linear systems.

Visual odometry

For visual odometry, people traditionally use filter-based methods, which can suffer from large linearization error. Nonlinear optimization-based methods have become more popular in recent years. These methods estimate the position and orientation of the robot by minimizing an error function, which is the difference between the re-projection of landmarks and their observation in the image frame. This procedure is called bundle adjustment because we are adjusting a bundle of light rays to match the projection with the observation.

fp16 SLAM.png
Bundle adjustment, in which “bundles” of light rays are adjusted to match projection with observation.

BA-based visual odometry operates over a sliding window that contains a fixed number of (key) frames. On average, a new key frame comes at 10Hz. The challenge is to solve the BA problem within a given time budget. One popular way to do this is to solve the normal equation that is the equivalent of the linearized optimization problem; this involves the approximation of the Hessian matrix, or the matrix of second-order derivatives of the residual.

Sparsity pattern.png
Sparsity patterns of Hessian matrices from bundle adjustment (left) and pose graph optimization (right).

The BA problem involves two sets of unknown state variables: one indicates the robot’s pose and the other indicates the landmark location. One way to reduce the computational burden of the BA problem is to marginalize the constraints between camera poses and landmarks and focus on the camera poses first. In the SLAM community, this procedure is known as Schur elimination or landmark marginalization.

Related content
Measuring the displacement between location estimates derived from different camera views can help enforce the local consistency vital to navigation.

This marginalization step can greatly reduce the size of the linear system that needs to be solved. For a 50-frame BA problem, the Jacobian matrix is usually of the size 5,500 x 1,000, and the Hessian is of size 1,000 x 1,000. Decoupling constraints reduces the size of the linear system to 300 x 300, small enough to be solved with direct or iterative solvers. However, this strategy requires both the formulation of the Hessian matrix and a partial-elimination step, which are expensive to employ in practice.

Our mixed-precision linear solver, which mixes single and half-precision, is based on the conjugate gradient normal-equation residual (CGNR) method, which is an iterative method directly applied to the linear-optimization problem without explicit formulation of the Hessian.

As in the general framework, a naïve casting of all computations to half-precision will result in lower accuracy. In our experiments, we found that if we compute matrix-vector products in half-precision and all other operations in single precision, we will maintain the overall accuracy of the SLAM pipeline.

Solver comparison.png
A comparison of the naïve half-precision solver (left) and the mixed-precision solver (right) on a single trajectory estimation.
Histogram.png
The cumulative-error histogram for 1,703 trajectory estimations where the VO is solved with mixed precision, half-precision, and double precision, respectively.

The matrix-vector products, which are the major computation in CGNR iterations, usually account for 83% of the computing cost, in terms of number of floating-point operations. That means that, if run on NVIDIA V100 GPUs, the mixed-precision solver could save at least 41% solving time compared to the single-precision linear solver.

Loop closure

In the SLAM pipeline, the local pose estimates from VO usually exhibit large drift, especially in the long run. Loop closure corrects this drift.

Loop closure.png
Illustration of loop closure.

For a real-world mapping estimate, without LC correction, the average trajectory error could be at the order of 0.1 meter, which is not acceptable in practice. This error is reduced to 10-4 meters after applying LC corrections.

 ATE w/o LC (m)ATE with LC (m)
Max4.03E-015.83E-04
99%2.65E-015.71E-04
90%2.00E-015.57E-04
Mean9.72E-023.19E-04

The LC adjustment involves solving a global PGO problem. Like the BA problem, it is a nonlinear optimization problem and can be solved within the same mixed-precision framework. But the linear systems arising from PGO problems are much larger and sparser than those of the BA problem.

Related content
“Body language” and an awareness of social norms help Amazon’s new household robot integrate gracefully into the home.

As more and more loops are closed, the problem size could grow from several hundreds of poses to several thousands of poses. If we measure the size of a matrix by the number of its rows, during loop closure, the size could grow from the order of 100 to the order of 10,000. Directly solving sparse matrices of this size in double precision is challenging, especially considering the time and computation constraints of on-device applications. For a real-world trajectory estimation, the solving time for the PGO problem could grow up to eight seconds with full CPU usage.

Solving times.png
Time for solving PGO problems during trajectory estimation. The x-axis represents the total number of key frames in each pose graph, and the y-axis represents the time for solving each PGO problem.

This results in a different strategy for designing a mixed-precision solver for PGO problems. Due to the sparsity of the Jacobian matrix, our mixed-precision method is still based on the iterative CGNR method. But to accelerate the convergence of the CGNR iterations, we apply a static incomplete Cholesky preconditioner in each iteration. Cholesky factorization decomposes a symmetric linear system into a product of two triangular matrices, meaning that all of their nonzero values are concentrated on one side of a diagonal across the matrix. This decomposition step is expensive, so we do it only once for the whole problem. The computational cost is mostly dominated by the application of the preconditioner, which involves solving two triangular systems. In our timing analysis, this step consumes around 50% of the computation in each linear solving.

To accelerate the optimization, instead of computing matrix-vector products in half-precision, we solve the triangular system in half-precision, keeping all other operations in single precision. With this mixed-precision solver, we could almost match the accuracy of the full-precision solver while reducing computing time by 26% on average.

ATE histogram
Cumulative ATE histogram for solving 800 PGO problems from a real-world trajectory estimation. Each PGO problem is solved with a mixed-precision solver and a single-precision solver, respectively.

Our results across both the VO and LC applications show that because of the high-efficiency and low-energy nature of half-precision arithmetic, mixed-precision solvers could make on-device SLAM faster and greener.

Acknowledgments

The following contributed equally to this work: Tong Qin, applied scientist, Amazon Hardware; Sankalp Dayal, applied-science manager, Hardware; Joydeep Biswas, software development engineer, Amazon Devices; Varada Gopalakrishnan, vice president and distinguished engineer, Hardware; Adam Fineberg, senior principal engineer, Devices; Rahul Bakshi, senior manager of software, machine learning, and mobility, Hardware.

Research areas

Related content

GB, MLN, Edinburgh
We’re looking for a Machine Learning Scientist in the Personalization team for our Edinburgh office experienced in generative AI and large models. You will be responsible for developing and disseminating customer-facing personalized recommendation models. This is a hands-on role with global impact working with a team of world-class engineers and scientists across the Edinburgh offices and wider organization. You will lead the design of machine learning models that scale to very large quantities of data, and serve high-scale low-latency recommendations to all customers worldwide. You will embody scientific rigor, designing and executing experiments to demonstrate the technical efficacy and business value of your methods. You will work alongside a science team to delight customers by aiding in recommendations relevancy, and raise the profile of Amazon as a global leader in machine learning and personalization. Successful candidates will have strong technical ability, focus on customers by applying a customer-first approach, excellent teamwork and communication skills, and a motivation to achieve results in a fast-paced environment. Our position offers exceptional opportunities for every candidate to grow their technical and non-technical skills. If you are selected, you have the opportunity to make a difference to our business by designing and building state of the art machine learning systems on big data, leveraging Amazon’s vast computing resources (AWS), working on exciting and challenging projects, and delivering meaningful results to customers world-wide. Key job responsibilities Develop machine learning algorithms for high-scale recommendations problems. Rapidly design, prototype and test many possible hypotheses in a high-ambiguity environment, making use of both quantitative analysis and business judgement. Collaborate with software engineers to integrate successful experimental results into large-scale, highly complex Amazon production systems capable of handling 100,000s of transactions per second at low latency. Report results in a manner which is both statistically rigorous and compellingly relevant, exemplifying good scientific practice in a business environment.
IN, TS, Hyderabad
Welcome to the Worldwide Returns & ReCommerce team (WWR&R) at Amazon.com. WWR&R is an agile, innovative organization dedicated to ‘making zero happen’ to benefit our customers, our company, and the environment. Our goal is to achieve the three zeroes: zero cost of returns, zero waste, and zero defects. We do this by developing products and driving truly innovative operational excellence to help customers keep what they buy, recover returned and damaged product value, keep thousands of tons of waste from landfills, and create the best customer returns experience in the world. We have an eye to the future – we create long-term value at Amazon by focusing not just on the bottom line, but on the planet. We are building the most sustainable re-use channel we can by driving multiple aspects of the Circular Economy for Amazon – Returns & ReCommerce. Amazon WWR&R is comprised of business, product, operational, program, software engineering and data teams that manage the life of a returned or damaged product from a customer to the warehouse and on to its next best use. Our work is broad and deep: we train machine learning models to automate routing and find signals to optimize re-use; we invent new channels to give products a second life; we develop highly respected product support to help customers love what they buy; we pilot smarter product evaluations; we work from the customer backward to find ways to make the return experience remarkably delightful and easy; and we do it all while scrutinizing our business with laser focus. You will help create everything from customer-facing and vendor-facing websites to the internal software and tools behind the reverse-logistics process. You can develop scalable, high-availability solutions to solve complex and broad business problems. We are a group that has fun at work while driving incredible customer, business, and environmental impact. We are backed by a strong leadership group dedicated to operational excellence that empowers a reasonable work-life balance. As an established, experienced team, we offer the scope and support needed for substantial career growth. Amazon is earth’s most customer-centric company and through WWR&R, the earth is our customer too. Come join us and innovate with the Amazon Worldwide Returns & ReCommerce team!
US, WA, Seattle
Prime Video is a first-stop entertainment destination offering customers a vast collection of premium programming in one app available across thousands of devices. Prime members can customize their viewing experience and find their favorite movies, series, documentaries, and live sports – including Amazon MGM Studios-produced series and movies; licensed fan favorites; and programming from Prime Video add-on subscriptions such as Apple TV+, Max, Crunchyroll and MGM+. All customers, regardless of whether they have a Prime membership or not, can rent or buy titles via the Prime Video Store, and can enjoy even more content for free with ads. Are you interested in shaping the future of entertainment? Prime Video's technology teams are creating best-in-class digital video experience. As a Prime Video technologist, you’ll have end-to-end ownership of the product, user experience, design, and technology required to deliver state-of-the-art experiences for our customers. You’ll get to work on projects that are fast-paced, challenging, and varied. You’ll also be able to experiment with new possibilities, take risks, and collaborate with remarkable people. We’ll look for you to bring your diverse perspectives, ideas, and skill-sets to make Prime Video even better for our customers. With global opportunities for talented technologists, you can decide where a career Prime Video Tech takes you! In Prime Video READI, our mission is to automate infrastructure scaling and operational readiness. We are growing a team specialized in time series modeling, forecasting, and release safety. This team will invent and develop algorithms for forecasting multi-dimensional related time series. The team will develop forecasts on key business dimensions with optimization recommendations related to performance and efficiency opportunities across our global software environment. As a founding member of the core team, you will apply your deep coding, modeling and statistical knowledge to concrete problems that have broad cross-organizational, global, and technology impact. Your work will focus on retrieving, cleansing and preparing large scale datasets, training and evaluating models and deploying them to production where we continuously monitor and evaluate. You will work on large engineering efforts that solve significantly complex problems facing global customers. You will be trusted to operate with complete independence and are often assigned to focus on areas where the business and/or architectural strategy has not yet been defined. You must be equally comfortable digging in to business requirements as you are drilling into design with development teams and developing production ready learning models. You consistently bring strong, data-driven business and technical judgment to decisions. You will work with internal and external stakeholders, cross-functional partners, and end-users around the world at all levels. Our team makes a big impact because nothing is more important to us than delivering for our customers, continually earning their trust, and thinking long term. You are empowered to bring new technologies to your solutions. If you crave a sense of ownership, this is the place to be.
US, WA, Seattle
Amazon Advertising operates at the intersection of eCommerce and advertising, and is investing heavily in building a world-class advertising business. We are defining and delivering a collection of self-service performance advertising products that drive discovery and sales. 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 to improve both shopper and advertiser experience. With a broad mandate to experiment and innovate, we grow at an unprecedented rate with a seemingly endless range of new opportunities. The Ad Response Prediction team in Sponsored Products organization build advanced deep-learning models, large-scale machine-learning pipelines, and real-time serving infra to match shoppers’ intent to relevant ads on all devices, for all contexts and in all marketplaces. Through precise estimation of shoppers’ interaction with ads and their long-term value, we aim to drive optimal ads allocation and pricing, and help to deliver a relevant, engaging and delightful ads experience to Amazon shoppers. As the business and the complexity of various new initiatives we take continues to grow, we are looking for talented Applied Scientists to join the team. Key job responsibilities As a Applied Scientist II, you will: * Conduct hands-on data analysis, build large-scale machine-learning models and pipelines * Work closely with software engineers on detailed requirements, technical designs and implementation of end-to-end solutions in production * Run regular A/B experiments, gather data, perform statistical analysis, and communicate the impact to senior management * Establish scalable, efficient, automated processes for large-scale data analysis, machine-learning model development, model validation and serving * Provide technical leadership, research new machine learning approaches to drive continued scientific innovation * Be a member of the Amazon-wide Machine Learning Community, participating in internal and external MeetUps, Hackathons and Conferences
US, WA, Bellevue
mmPROS Surface Research Science seeks an exceptional Applied Scientist with expertise in optimization and machine learning to optimize Amazon's middle mile transportation network, the backbone of its logistics operations. Amazon's middle mile transportation network utilizes a fleet of semi-trucks, trains, and airplanes to transport millions of packages and other freight between warehouses, vendor facilities, and customers, on time and at low cost. The Surface Research Science team delivers innovation, models, algorithms, and other scientific solutions to efficiently plan and operate the middle mile surface (truck and rail) transportation network. The team focuses on large-scale problems in vehicle route planning, capacity procurement, network design, forecasting, and equipment re-balancing. Your role will be to build innovative optimization and machine learning models to improve driver routing and procurement efficiency. Your models will impact business decisions worth billions of dollars and improve the delivery experience for millions of customers. You will operate as part of a team of innovative, experienced scientists working on optimization and machine learning. You will work in close collaboration with partners across product, engineering, business intelligence, and operations. Key job responsibilities - Design and develop optimization and machine learning models to inform our hardest planning decisions. - Implement models and algorithms in Amazon's production software. - Lead and partner with product, engineering, and operations teams to drive modeling and technical design for complex business problems. - Lead complex modeling and data analyses to aid management in making key business decisions and set new policies. - Write documentation for scientific and business audiences. About the team This role is part of mmPROS Surface Research Science. Our mission is to build the most efficient and optimal transportation network on the planet, using our science and technology as our biggest advantage. We leverage technologies in optimization, operations research, and machine learning to grow our businesses and solve Amazon's unique logistical challenges. Scientists in the team work in close collaboration with each other and with partners across product, software engineering, business intelligence, and operations. They regularly interact with software engineering teams and business leadership.
IL, Tel Aviv
Come join the AWS Agentic AI science team in building the next generation models for intelligent automation. AWS, the world-leading provider of cloud services, has fostered the creation and growth of countless new businesses, and is a positive force for good. Our customers bring problems that will give Applied Scientists like you endless opportunities to see your research have a positive and immediate impact in the world. You will have the opportunity to partner with technology and business teams to solve real-world problems, have access to virtually endless data and computational resources, and to world-class engineers and developers that can help bring your ideas into the world. As part of the team, we expect that you will develop innovative solutions to hard problems, and publish your findings at peer reviewed conferences and workshops. We are looking for world class researchers with experience in one or more of the following areas - autonomous agents, API orchestration, Planning, large multimodal models (especially vision-language models), reinforcement learning (RL) and sequential decision making. Key job responsibilities PhD, or Master's degree and 4+ years of CS, CE, ML or related field experience 3+ years of building models for business application experience Experience in patents or publications at top-tier peer-reviewed conferences or journals Experience programming in Java, C++, Python or related language Experience in any of the following areas: algorithms and data structures, parsing, numerical optimization, data mining, parallel and distributed computing, high-performance computing
IL, Haifa
Come join the AWS Agentic AI science team in building the next generation models for intelligent automation. AWS, the world-leading provider of cloud services, has fostered the creation and growth of countless new businesses, and is a positive force for good. Our customers bring problems that will give Applied Scientists like you endless opportunities to see your research have a positive and immediate impact in the world. You will have the opportunity to partner with technology and business teams to solve real-world problems, have access to virtually endless data and computational resources, and to world-class engineers and developers that can help bring your ideas into the world. As part of the team, we expect that you will develop innovative solutions to hard problems, and publish your findings at peer reviewed conferences and workshops. We are looking for world class researchers with experience in one or more of the following areas - autonomous agents, API orchestration, Planning, large multimodal models (especially vision-language models), reinforcement learning (RL) and sequential decision making.
US, NY, New York
Join us in a historic endeavor to make Generative AI accessible to the world with breakthrough research! The AWS AI team has a world-leading team of researchers and academics, and we are looking for world-class colleagues to join us and make the AI revolution happen. Our team of scientists drives the innovation that enables external and internal SageMaker customers to train their next generation models on both GPU and Trainium instances. As part of the team, we expect that you will develop innovative solutions to hard problems, and publish your findings at peer reviewed conferences and workshops. AWS is the world-leading provider of cloud services, has fostered the creation and growth of countless new businesses, and is a positive force for good. Our customers bring problems which will give Applied Scientists like you endless opportunities to see your research have a positive and immediate impact in the world. You will have the opportunity to partner with technology and business teams to solve real-world problems, have access to virtually endless data and computational resources, and to world-class engineers and developers that can help bring your ideas into the world. About the team Why AWS Amazon Web Services (AWS) is the world’s most comprehensive and broadly adopted cloud platform. We pioneered cloud computing and never stopped innovating — that’s why customers from the most successful startups to Global 500 companies trust our robust suite of products and services to power their businesses. Utility Computing (UC) AWS Utility Computing (UC) provides product innovations — from foundational services such as Amazon’s Simple Storage Service (S3) and Amazon Elastic Compute Cloud (EC2), to consistently released new product innovations that continue to set AWS’s services and features apart in the industry. As a member of the UC organization, you’ll support the development and management of Compute, Database, Storage, Internet of Things (IoT), Platform, and Productivity Apps services in AWS, including support for customers who require specialized security solutions for their cloud services. Inclusive Team Culture Here at AWS, it’s in our nature to learn and be curious. Our employee-led affinity groups foster a culture of inclusion that empower us to be proud of our differences. Ongoing events and learning experiences, including our Conversations on Race and Ethnicity (CORE) and AmazeCon (gender diversity) conferences, inspire us to never stop embracing our uniqueness. Work/Life Balance We value work-life harmony. Achieving success at work should never come at the expense of sacrifices at home, which is why we strive for flexibility as part of our working culture. When we feel supported in the workplace and at home, there’s nothing we can’t achieve in the cloud. Mentorship and Career Growth We’re continuously raising our performance bar as we strive to become Earth’s Best Employer. That’s why you’ll find endless knowledge-sharing, mentorship and other career-advancing resources here to help you develop into a better-rounded professional. Diverse Experiences Amazon values diverse experiences. Even if you do not meet all of the preferred qualifications and skills listed in the job description, we encourage candidates to apply. If your career is just starting, hasn’t followed a traditional path, or includes alternative experiences, don’t let it stop you from applying.
US, WA, Seattle
Prime Video is a first-stop entertainment destination offering customers a vast collection of premium programming in one app available across thousands of devices. Prime members can customize their viewing experience and find their favorite movies, series, documentaries, and live sports – including Amazon MGM Studios-produced series and movies; licensed fan favorites; and programming from Prime Video add-on subscriptions such as Apple TV+, Max, Crunchyroll and MGM+. All customers, regardless of whether they have a Prime membership or not, can rent or buy titles via the Prime Video Store, and can enjoy even more content for free with ads. Are you interested in shaping the future of entertainment? Prime Video's technology teams are creating best-in-class digital video experience. As a Prime Video team member, you’ll have end-to-end ownership of the product, user experience, design, and technology required to deliver state-of-the-art experiences for our customers. You’ll get to work on projects that are fast-paced, challenging, and varied. You’ll also be able to experiment with new possibilities, take risks, and collaborate with remarkable people. We’ll look for you to bring your diverse perspectives, ideas, and skill-sets to make Prime Video even better for our customers. With global opportunities for talented technologists, you can decide where a career Prime Video Tech takes you! Key job responsibilities As an Applied Scientist in the Content Understanding Team, you will lead the end-to-end research and deployment of video and multi-modal models applied to a variety of downstream applications. More specifically, you will: - Work backwards from customer problems to research and design scientific approaches for solving them - Work closely with other scientists, engineers and product managers to expand the depth of our product insights with data, create a variety of experiments to determine the high impact projects to include in planning roadmaps - Stay up-to-date with advancements and the latest modeling techniques in the field - Publish your research findings in top conferences and journals About the team Our Prime Video Content Understanding team builds holistic media representations (e.g. descriptions of scenes, semantic embeddings) and apply them to new customer experiences supply chain problems. Our technology spans the entire Prime Video catalogue globally, and we enable instant recaps, skip intro timing, ad placement, search, and content moderation.
US, MA, Boston
The Artificial General Intelligence (AGI) team is looking for a passionate, talented, and inventive Applied Scientist with a strong deep learning background, to build industry-leading technology with Large Language Models (LLMs) and multi-modal systems. You will support projects that work on technologies including multi-modal model alignment, moderation systems and evaluation. Key job responsibilities As an Applied Scientist with the AGI team, you will support the development of novel algorithms and modeling techniques, to advance the state of the art with LLMs. 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 generative artificial intelligence (GenAI). You are also expected to publish in top tier conferences. About the team The AGI team has a mission to push the envelope in LLMs and multimodal systems. Specifically, we focus on model alignment with an aim to maintain safety while not denting utility, in order to provide the best-possible experience for our customers.