Social
Conversational AI

A quick guide to Amazon’s 50-plus papers at EMNLP 2024

Large language models predominate, both as a research subject themselves and as tools for researching topics of particular interest to Amazon, such as speech, recommendations, and information retrieval.

By Staff writer
November 14, 2024
7 min read
Share

Large language models (LLMs) have come to dominate the field of natural-language processing, so it’s no surprise that they also dominate the research that Amazon scientists are presenting at this year’s Conference on Empirical Methods in Natural-Language Processing (EMNLP). LLM training is the topic with the greatest number of Amazon papers, followed closely by strategies for mitigating misinformation in LLMs’ outputs — including but not limited to hallucinations. At the same time, a number of papers apply LLMs to topics of traditional interest at Amazon, such as speech, recommender systems, and information retrieval. (Papers marked with asterisks were accepted to Findings of EMNLP.)

AI agents

MARCO: Multi-agent real-time chat orchestration
Anubhav Shrimal, Shervin Malmasi, Kriti Biswas, Swarnalatha Raghuraman, Anish Nediyanchath, Yi Zhang, Promod Yenigalla

Code generation

CodeFort: Robust training for code generation models
Yuhao Zhang, Shiqi Wang, Haifeng Qian, Zijian Wang, Mingyue Shang, Linbo Liu, Sanjay Krishna Gouda, Baishakhi Ray, Murali Krishna Ramanathan, Xiaofei Ma, Anoop Deoras

Socratic human feedback (SoHF): Expert steering strategies for LLM code generation
Subramanian Chidambaram, Erran Li, Min Bai, Xiaopeng LI, Kaixiang Lin, Xiong Zhou, Alex C. Williams

Structured object language modeling (SoLM): Native structured objects generation conforming to complex schemas with self-supervised denoising
Amir Tavanaei, Kee Kiat Koo, Hayreddin Ceker, Shaobai Jiang, Qi Li, Julien Han, Karim Bouyarmane

Contrastive decoding

Explaining and improving contrastive decoding by extrapolating the probabilities of a huge and hypothetical LM
Haw-Shiuan Chang, Nanyun Peng, Mohit Bansal, Anil Ramakrishna, Tagyoung Chung

Explaining and improving contrastive decoding by extrapolating the probabilities of a huge and hypothetical LM.png
Given a simple question with clues, contrastive decoding could have an “obvious blindness” (e.g., assigning higher probability to an uncommon answer, such as "invertebrate", than to the most obvious answer, "bees"). In contrast, the asymptotic probability decoding proposed in "Explaining and improving contrastive decoding by extrapolating the probabilities of a huge and hypothetical LM" correctly assigns the highest probability to "bees" by leveraging the probabilities from multiple LMs of different sizes.

Data integration

ASTRA: Automatic schema matching using machine translation
Tarang Chugh, Deepak Zambre

Learning from natural language explanations for generalizable entity matching
Somin Wadhwa, Adit Krishnan, Runhui Wang, Byron C. Wallace, Chris (Luyang) Kong

Pretraining and finetuning language models on geospatial networks for accurate address matching
Saket Maheshwary, Arpan Paul, Saurabh Sohoney

Retrieval augmented spelling correction for e-commerce applications
Xuan Guo, Rohit Patki, Dante Everaert, Christopher Potts

Dataset distillation

Textual dataset distillation via language model embedding
Yefan Tao, Chris (Luyang) Kong, Andrey Kan, Laurent Callot

Textual dataset distillation via language model embedding: DaLLME.png
The DaLLME framework proposed in "Textual dataset distillation via language model embedding" begins by using a language model to transform raw textual data into embedding vectors. A set of distilled vectors is then derived in the embedding space, through a process designed to encapsulate maximum informational content. Finally, the vec2text model translates these distilled vectors back into textual form.

Document understanding

DocKD: Knowledge distillation from LLMs for open-world document understanding models
Sungnyun Kim, Haofu Liao, Srikar Appalaraju, Peng Tang, Zhuowen Tu, Ravi Kumar Satzoda, R. Manmatha, Vijay Mahadevan, Stefano Soatto

Information retrieval

Evaluating D-MERIT of partial-annotation on information retrieval
Royi Rassin, Yaron Fairstein, Oren Kalinsky, Guy Kushilevitz, Nachshon Cohen, Alexander Libov, Yoav Goldberg

Identifying high consideration e-commerce search queries
Zhiyu Chen, Jason Choi, Besnik Fetahu, Shervin Malmasi

Learning when to retrieve, what to rewrite, and how to respond in conversational QA*
Nirmal Roy, Leonardo Ribeiro, Rexhina Blloshmi, Kevin Small

Natural-language understanding

Intent detection in the age of LLMs
Gaurav Arora, Shreya Jain, Srujana Merugu

Intent detection in the age of LLMs.png
"Intent detection in the age of LLMs" proposes a methodology for adaptive in-context learning and chain-of-thought-based intent detection using LLMs.

Predicting entity salience in extremely short documents
Ben Bullough, Harrison Lundberg, Chen Hu, Weihang Xiao

LLM evaluation

AXCEL: Automated eXplainable consistency evaluation using LLMs*
P Aditya Sreekar, Sahil Verma, Suransh Chopra, Sarik Ghazarian, Abhishek Persad, Narayanan Sadagopan

Precise model benchmarking with only a few observations
Riccardo Fogliato, Pratik Patil, Nil-Jana Akpinar, Mathew Monfort

LLM fine tuning

AdaZeta: Adaptive zeroth-order tensor-train adaption for memory-efficient large language models fine-tuning
Yifan Yang, Kai Zhen, Ershad Banijamali, Thanasis Mouchtaris, Zheng Zhang

RoseLoRA: Row and column-wise sparse low-rank adaptation of pre-trained language model for knowledge editing and fine-tuning
Haoyu Wang, Tianci Liu, Ruirui Li, Monica Cheng, Tuo Zhao, Jing Gao

RoseLoRA.png
The row- and column-wise sparse low-rank adaptation (RoseLoRA) framework proposed in "RoseLoRA: Row and column-wise sparse low-rank adaptation of pre-trained language model for knowledge editing and fine-tuning".

LLMs for speech

Speechworthy instruction-tuned language models
Hyundong Cho, Nicolaas Jedema, Leonardo Ribeiro, Karishma Sharma, Pedro Szekely, Alessandro Moschitti, Ruben Janssen, Jonathan May

LLM misinformation mitigation

ECON: On the detection and resolution of evidence conflicts
Cheng Jiayang, Chunkit Chan, Qianqian Zhuang, Lin Qiu, Tianhang Zhang, Tengxiao Liu, Yangqiu Song, Yue Zhang, Pengfei Liu, Zheng Zhang

Generative subgraph retrieval for knowledge graph–grounded dialog generation
Jinyoung Park, Minseok Joo, Joo-Kyung Kim, Hyunwoo J. Kim

HalluMeasure: Fine-grained hallucination measurement using chain-of-thought reasoning
Shayan Ali Akbar, Md Mosharaf Hossain, Tess Wood, Si-Chi Chin, Erica Salinas, Victor Alvarez, Erwin Cornejo

Knowledge-centric hallucination detection
Xiangkun Hu, Dongyu Ru, Lin Qiu, Qipeng Guo, Tianhang Zhang, Yang Xu, Yun Luo, Pengfei Liu, Zheng Zhang, Yue Zhang

LLM reasoning

Auto-evolve: Enhancing large language model’s performance via self-reasoning framework*
Krishna Aswani, Alex Lu, Pranav Patankar, Priya Dhalwani, Iris Tan, Jayant Ganeshmohan, Simon Lacasse

LLM self-correction

LLM self-correction with DeCRIM: Decompose, critique, and refine for enhanced following of instructions with multiple constraints
Thomas Palmeira Ferraz, Kartik Mehta, Yu-Hsiang Lin, Haw-Shiuan Chang, Shereen Oraby, Sijia Liu, Vivek Subramanian, Tagyoung Chung, Mohit Bansal, Nanyun Peng

DeCRIM.png
In the DeCRIM pipeline proposed in "LLM self-correction with DeCRIM: Decompose, critique, and refine for enhanced following of instructions with multiple constraints", an LLM first generates a response to a user request. The Decomposer then breaks down the request into granular constraints, and the Critic model gives feedback on whether the response meets those constraints. If it does, the response is output; if not, the LLM uses the feedback to refine the response.

LLM training

Dancing in chains: Reconciling instruction following and faithfulness in language models
Zhengxuan Wu, Yuhao Zhang, Peng Qi, Yumo Xu, Rujun Han, Yian Zhang, Jifan Chen, Bonan Min, Zhiheng Huang

DEM: Distribution edited model for training with mixed data distributions
Dhananjay Ram, Aditya Rawal, Momchil Hardalov, Nikolaos Pappas, Sheng Zha

DEM: Distribution Edited Model for Training with Mixed Data Distributions
The distribution-edited model (ΘD) described in "DEM: Distribution edited model for training with mixed data distributions" results from fine-tuning a pretrained model (Θ) on n individual data distributions (Di) and combining the resulting models with basic element-wise vector operations. Here, the extracted distribution vectors (∆ΘDi ) are multiplied by weight coefficients, and the weighted sum is added to the base model.

Evolutionary contrastive distillation for language model alignment
Julian Katz-Samuels, Zheng Li, Hyokun Yun, Priyanka Nigam, Yi Xu, Vaclav Petricek, Bing Yin, Trishul Chilimbi

Hop, skip, jump to convergence: Dynamics of learning rate transitions for improved training of large language models
Shreyas Subramanian, Vignesh Ganapathiraman, Corey Barrett

Learning from relevant subgoals in successful dialogs using iterative training for task-oriented dialog systems
Magdalena Kaiser, Patrick Ernst, Gyuri Szarvas

Quality matters: Evaluating synthetic data for tool-using LLMs
Shadi Iskander, Nachshon Cohen, Zohar Karnin, Ori Shapira, Sofia Tolmach

Query autocompletion

AmazonQAC: A large-scale, naturalistic query autocomplete dataset
Dante Everaert, Rohit Patki, Tianqi Zheng, Christopher Potts

DiAL: Diversity aware listwise ranking for query auto-complete
Sonali Singh, Sachin Farfade, Prakash Mandayam Comar

Question answering

RAG-QA arena: Evaluating domain robustness for long-form retrieval-augmented question answering
Rujun Han, Yuhao Zhang, Peng Qi, Yumo Xu, Jenyuan Wang, Lan Liu, William Yang Wang, Bonan Min, Vittorio Castelli

Retrieving contextual information for long-form question answering using weak supervision
Philipp Christmann, Svitlana Vakulenko, Ionut Teodor Sorodoc, Bill Byrne, Adrià de Gispert

Recommender systems

Efficient pointwise-pairwise learning-to-rank for news recommendation
Nithish Kannen Senthilkumar, Yao Ma, Gerrit van den Burg, Jean Baptiste Faddoul

Efficient pointwise-pairwise learning-to-rank for news recommendation.png
An illustration of the GLIMPSE framework proposed in "Efficient pointwise-pairwise learning-to-rank for news recommendation". GLIMPSE adopts a multitask approach in which a pretrained language model is fine-tuned on both the relevance prediction task and the pairwise-preference task. During inference, the relevance predictions are used to produce an initial pointwise ranking, which is subsequently improved by one or more right-to-left (RTL) passes using pairwise comparisons.

PEARL: Preference extraction with exemplar augmentation and retrieval with LLM agents
Vijit Malik, Akshay Jagatap, Vinayak Puranik, Anirban Majumder

Sequential LLM framework for fashion recommendation
Han Liu, Xianfeng Tang, Tianlang Chen, Jiapeng Liu, Indu Indu, Henry Peng Zou, Peng Dai, Roberto Fernandez Galan, Mike Porter, Dongmei Jia, Ning Zhang, Lian Xiong

Responsible AI

Attribute controlled fine-tuning for large language models: A case study on detoxification
Tao Meng, Ninareh Mehrabi, Palash Goyal, Anil Ramakrishna, Aram Galstyan, Richard Zemel, Kai-Wei Chang, Rahul Gupta, Charith Peris

FLIRT: Feedback loop in-context red teaming
Ninareh Mehrabi, Palash Goyal, Christophe Dupuy, Qian Hu, Shalini Ghosh, Richard Zemel, Kai-Wei Chang, Aram Galstyan, Rahul Gupta

Order of magnitude speedups for LLM membership inference
Rongting Zhang, Martin Bertran Lopez, Aaron Roth

Synthetic data generation

CorrSynth: A correlated sampling method for diverse dataset generation from LLMs
Suhas Kowshik, Abhishek Divekar, Vijit Malik

A Correlated Sampling Method for Diverse Dataset Generation from LLMs
"CorrSynth: A correlated sampling method for diverse dataset generation from LLMs" introduces a sampling method that uses anti-correlation between examples rather than few-shot generation.

DATA ADVISOR: Dynamic data curation for safety alignment of large language models
Fei Wang, Ninareh Mehrabi, Palash Goyal, Rahul Gupta, Kai-Wei Chang, Aram Galstyan

Evaluating differentially private synthetic data generation in high-stakes domains
Krithika Ramesh, Nupoor Gandhi, Pulkit Madaan, Lisa Bauer, Charith Peris, Anjalie Field

SYNTHESIZRR: Generating diverse datasets with retrieval augmentation
Abhishek Divekar, Greg Durrett

Abstract depiction of the SYNTHESIZRR procedure
Abstract depiction of the procedure proposed in "SYNTHESIZRR: Generating diverse datasets with retrieval augmentation". The content sourcing stage retrieves K unique documents {r1,...,rK} from a large corpus for each in-context covariate xICL. The task-inversion stage uses a parameterized context refinement prompt, Pτ, which takes parameters Rinv (inversion instruction), rk (a retrieved document), and V(yICL) (the verbalized target label). A generalist teacher LLM autoregressively generates a synthetic covariate. Each in-context example thus produces K unique synthetic examples {x̃1,..., x̃K}, which we include in the dataset with target yICL.

Text classification

Distance-aware calibration for pre-trained language models*
Alberto Gasparin, Gianluca Detommaso

Performance-guided LLM knowledge distillation for efficient text classification at scale

Flavio Di Palo, Prateek Singhi, Bilal Fadlallah

Prompt-tuned muti-task taxonomic transformer (PTMTTaxoFormer)
Rajashekar Vasantha, Nhan Nguyen, Yue Zhang

Text summarization

Salient information prompting to steer content in prompt-based abstractive summarization
Lei Xu, Asad Karim, Saket Dingliwal, Aparna Elangovan

Research areas
Tags
About the Author
Staff writer

Related content

Work with us

See more jobs See more jobs
Manager, Applied Science- Manipulation & Autonomy , Fauna
US, NY, New York
We are seeking a Lead Applied Scientist to drive the development of next-generation manipulation and autonomy systems for robots operating in complex, real-world environments. This role is designed for an exceptional technical leader with a strong research background and a passion for translating cutting-edge ideas into working robotic systems. You will play a central role in defining the technical direction of our manipulation and autonomy stack, from problem formulation and algorithm design to system integration and real-world deployment. While this role is hands-on and deeply technical, it is also expected to evolve toward technical leadership and team building over time. Key job responsibilities - Lead research and development of manipulation and autonomy systems for robots, including planning, control, learning, and closed-loop execution - Design and implement algorithms, especially focused around autonomy and interaction with dynamic environments - Work closely with perception, motor control, hardware, and systems teams to build tightly integrated autonomy pipelines - Define evaluation methodologies, benchmarks, and simulation-to-real workflows for manipulation and autonomy - Stay deeply engaged with the research community, selectively incorporating state-of-the-art techniques into deployed systems - Mentor and guide other researchers and engineers, help set technical direction, and contribute to team growth and research culture - Lead technical projects from conception through production deployment - Bridge research initiatives with practical engineering implementation About the team Fauna Robotics, an Amazon company, is building capable, safe, and genuinely delightful robots for everyday life. Our goal is simple: make robots people actually want to live and interact with in everyday human spaces. We believe that future won’t arrive until building for robotics becomes far more accessible. Today, too much effort is spent reinventing the fundamentals. We’re changing that by developing tightly integrated hardware and software systems that make it faster, safer, and more intuitive to create real-world robotic products. Our work spans the full stack: mechanical design, control systems, dynamic modeling, and intelligent software. The focus is not just functionality, but experience. We’re building robots that feel responsive, expressive, and genuinely useful. At Fauna, you’ll work at the frontier of this space, helping define how robots move, manipulate, and interact with people in natural environments. It’s an opportunity to solve hard problems across hardware and software with a team focused on making robotics accessible and joyful to build. If you care about making robotics real for everyone and building systems that are as delightful as they are capable, we’re interested in hearing from you.
Applied Scientist, AWS Applied AI Solutions - Life Sciences
US, WA, Seattle
As part of the AWS Applied AI Solutions organization, we have a vision to provide end user applications, leveraging Amazon's unique experience and expertise, that are used by millions of companies worldwide to manage day-to-day operations. We will accomplish this by accelerating our customers' businesses through delivery of intuitive and differentiated technology solutions that solve enduring business challenges. We blend vision with curiosity and Amazon's real-world experience to build opinionated, turnkey solutions. Where customers prefer to buy over build, we become their trusted partner with solutions that are easy to adopt and easy to use. The Team Join the next science revolution at AWS Life Sciences Applied AI Solutions, where you'll work alongside world-class scientists to build AI that transforms how therapeutics are discovered, developed, and brought to patients. We're out to revolutionize how medicines are discovered, developed, and brought to patients, powered by a new generation of AI. Our team tackles some of the hardest open problems at the intersection of frontier AI and life sciences. We apply biological foundation models, large language models, and agentic reasoning systems to life sciences problems, then put them into the hands of customers as applications and managed services they can fine-tune, tailor, and deploy on their own data. The science challenges are deep: how do you design agentic systems that reason correctly over complex biological, regulatory, and clinical logic? How do you enable customers to tailor foundation models to their proprietary data and get better outputs with less effort? How do you adapt models to reason faithfully in high-stakes scientific and regulatory domains? Today we're focused on two areas. In clinical trials, we're building AI that automates and optimizes regulatory and clinical development workflows. In drug design, our products (including Amazon Bio Discovery) accelerate discovery by giving bench scientists AI-guided protein engineering and antibody design capabilities. We combine frontier research with production-scale delivery to put breakthrough science into the hands of customers solving humanity's hardest problems. We value scientific rigor, encourage publication, and support conference participation. If you want to do research that ships, this is the team. The Role We are seeking an Applied Scientist to build the models and methods behind our life sciences AI products, with a primary focus on clinical trial operations and agentic reasoning. You will design, train, and evaluate systems that reason over complex clinical and operational logic, and ship them into products customers use directly. You will work closely with senior and principal scientists on well-scoped research problems, own your results end to end, and see your work reach production. This role combines expertise in LLM reasoning and agentic AI with applied impact in life sciences. You will work on how large language models reason, plan, and act in complex scientific domains, while applying domain knowledge to ensure models produce scientifically valid outputs. The problems span multiple fronts: - How do you build LLM-based agentic systems that correctly reason over clinical protocols, regulatory standards, and complex multi-step operational workflows? - How do you evaluate agent reliability and faithfulness rigorously enough to trust in high-stakes clinical settings? - How do you develop model customization methods (fine-tuning, retrieval augmentation, domain adaptation) that let customers get strong results from foundation models on their own data? You will focus on clinical trial operations (agentic automation, structured reasoning, evaluation, domain adaptation), with opportunities to contribute across drug discovery (protein engineering, antibody design) as the portfolio grows. You will own end-to-end scientific solutions from research through production, and your work will directly shape the tools that scientists use daily. Key job responsibilities - Design, train, fine-tune, and evaluate LLM-based agentic systems that reason over clinical protocols, regulatory standards, and operational workflows - Build rigorous evaluation harnesses and benchmarks to measure agent reliability, faithfulness, and failure modes in high-stakes domains - Develop model customization methods (fine-tuning, RLHF, retrieval augmentation, domain adaptation) that help customers get better outputs on their own data with less effort - Contribute to graph-based and causal modeling approaches for clinical trial operations - Partner with Life Sciences domain experts, product, and engineering to translate scientific challenges into shipped capabilities - Own experiments end to end: problem framing, implementation, evaluation, iteration, and hand-off to production - Publish at top-tier venues where the work supports it - Contribute to drug discovery efforts (protein engineering, antibody design) as opportunities arise A day in the life - Design and run an experiment to validate a new agentic reasoning or fine-tuning method, then ship it as a capability customers can use - Diagnose why a model is failing on a new class of inputs and implement a fix to unblock a delivery milestone - Build or extend an evaluation benchmark to measure how faithfully an agent reasons over clinical logic - Meet with domain experts to scope what the next model release needs to do - Review results with a senior scientist, sharpen the approach, and get it over the finish line - Prototype a new idea that could become the next capability in the product About the team AWS Solutions As part of the AWS solutions organization, we have a vision to provide business applications, leveraging Amazon's unique experience and expertise, that are used by millions of companies worldwide to manage day-to-day operations. We will accomplish this by accelerating our customers' businesses through delivery of intuitive and differentiated technology solutions that solve enduring business challenges. we blend vision with curiosity and Amazon's real-world experience to build opinionated, turnkey solutions. Where customers prefer to buy over build, we become their trusted partner with solutions that are no-brainers to buy and easy to use. Diverse Experiences AWS 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. 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. Inclusive Team Culture AWS values curiosity and connection. Our employee-led and company-sponsored affinity groups promote inclusion and empower our people to take pride in what makes us unique. Our inclusion events foster stronger, more collaborative teams. Our continual innovation is fueled by the bold ideas, fresh perspectives, and passionate voices our teams bring to everything we do. Mentorship & 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. 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.
Applied Scientist, Responsible AI
US, WA, Seattle
We are Amazon's central Responsible AI team. Our mission is to advance the science and practice of Responsible AI (RAI) to enable Amazon's tens of thousands of builders to build and deploy AI solutions to the high standards that our customers and society expect. As a scientist on this team, you will: - understand in depth the technical and scientific issues related to RAI, including controllability, security, privacy, safety, veracity, robustness, fairness, explainability, transparency, and governance - help define the strategies, priorities and metrics for RAI - solve open problems in RAI to unblock traditional, generative and/or agentic AI use cases and solutions, publishing as appropriate - help develop our team - liase with internal and external stakeholders, including the academic community, on issues related to RAI About the team Diverse Experiences Amazon Security values diverse experiences. Even if you do not meet all of the 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. Why Amazon Security? At Amazon, security is central to maintaining customer trust and delivering delightful customer experiences. Our organization is responsible for creating and maintaining a high bar for security across all of Amazon’s products and services. We offer talented security professionals the chance to accelerate their careers with opportunities to build experience in a wide variety of areas including cloud, devices, retail, entertainment, healthcare, operations, and physical stores. Inclusive Team Culture In Amazon Security, it’s in our nature to learn and be curious. Ongoing DEI events and learning experiences inspire us to continue learning and to embrace our uniqueness. Addressing the toughest security challenges requires that we seek out and celebrate a diversity of ideas, perspectives, and voices. Training & 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, training, and other career-advancing resources here to help you develop into a better-rounded professional. 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 flexible work hours and arrangements are part of our culture. When we feel supported in the workplace and at home, there’s nothing we can’t achieve.
Applied Scientist - Amazon Robotics, Amazon Robotics - Vulcan Stow
US, WA, Seattle
Our team in Amazon Robotics builds robotic systems that perform contact-rich manipulation tasks safely and reliably in complex, unstructured environments — at Amazon scale. Our scientists and engineers push the boundaries of robotic manipulation to handle enormous object diversity, bringing deep expertise across planning, control, perception, and machine learning. We learn from real-world data at a scale that few teams in robotics can access. We are seeking an Applied Scientist to join our Motion Behaviors team. You will drive the development of learned controllers and manipulation behaviors, applying techniques like reinforcement learning and behavior cloning to robots operating in Amazon fulfillment centers. These problems remain unsolved at scale: our robots must improve continuously in environments where simulation alone is insufficient. You will make principled decisions about when learned approaches should replace engineered solutions, and how to select behaviors based on estimated risk. You will collaborate across disciplines and leverage rich operational data to continuously improve system performance. Key job responsibilities • Develop learned controllers and manipulation behaviors, from research prototyping through deployment on production robots. • Research, design, and implement motion planning, control, and decision-making algorithms that improve the performance of deployed systems. • Design and deploy learning pipelines that take policies from simulation training to reliable, real-time execution on physical robots. • Develop models that predict manipulation outcomes and inform behavior selection under uncertainty. • Leverage operational data from deployed systems to systematically identify failure modes and drive policy improvements. • Represent Amazon in academia through publications and scientific presentations. A day in the life Amazon offers a full range of benefits that assist you and eligible family members, including domestic partners. Benefits can vary by location, the number of regularly scheduled hours you work, length of employment, and job status such as seasonal or temporary employment. The benefits that generally apply to regular, full-time employees include: 1. Medical, Dental, and Vision Coverage 2. Maternity and Parental Leave Options 3. Paid Time Off (PTO) 4. 401(k) Plan If you are not sure that every qualification on the list above describes you exactly, we'd still love to hear from you! At Amazon, we value people with unique backgrounds, experiences, and skillsets. If you’re passionate about this role and want to make an impact on a global scale, please apply!
Sr. Data Scientist, Amazon Pay Data Products
IN, KA, Bengaluru
Amazon Pay strives to be Earth’s most customer-centric payments service. Our mission is to serve customers and merchant partners with the most trusted, friction-less and rewarding payment solutions for their needs on and off Amazon. We are seeking an exceptional Data Scientist III to drive innovation in machine learning and artificial intelligence solutions while leading high-impact initiatives across the organization. Key job responsibilities Technical Excellence Lead end-to-end machine learning projects using PyTorch, AWS SageMaker, and other leading ML frameworks Design and implement complex statistical models and deep learning solutions Develop and optimize MLOps pipelines for model training, evaluation, and deployment Experience with modern LLM frameworks and Generative AI applications Expertise in Python, R, and related data science libraries MLOps & Development Build automated ML pipelines using AWS services (CodePipeline, Lambda, Step Functions) Implement CI/CD practices for ML model deployment and monitoring Create containerized solutions using Docker for scalable model deployment Experience with model optimization and hyperparameter tuning using tools like Optuna Integrate ML solutions with monitoring tools like MLflow Business Impact & Leadership Partner with stakeholders to translate business problems into technical solutions Design and develop business intelligence applications for real-time insights Lead technical initiatives and mentor junior data scientists Drive cross-functional collaboration to deliver innovative solutions Communicate complex technical concepts to non-technical audiences About the team The Amazon Pay Data Products team is a central unit that builds and maintains data products supporting Amazon Pay's growth across multiple markets. We operate at scale, processing 150M+ monthly transactions and managing 12 PB of data infrastructure. Our team consists of Business Intelligence Engineers, Data Engineers, and Product Managers who develop and maintain standardized reporting, data marts, and self-service analytics tools. Our expanded capabilities cover data science and Gen AI wherein we have built our first suite of multi-agent systems.
Principal Applied Scientist, AAIS
US, WA, Seattle
AI assistants are getting genuinely good at remembering individuals: your preferences, your projects, the thread you left open last week. But that memory stops at the edge of one person's usage. It doesn't reach the level at which real work happens, where the knowledge that matters is spread across many people, where one person's decision changes what everyone else should do next, and where nobody has the full picture. We're building AI that operates at that level: a durable, accurate understanding of how a team works, used to make that team measurably faster. We are looking for a Principal Applied Scientist to own the scientific direction of that work. This is a broad, ambiguous, high-leverage charter. The problems span knowledge representation, temporal reasoning, retrieval, agentic behavior, and the measurement science needed to know whether any of it is working. You will not be handed a well-posed problem. You will decide which problems are worth posing. This is a science leadership role, not a solo research role. You will set direction and raise the scientific bar across a team of applied scientists and MLEs, while staying deep enough in the work to prototype an idea yourself and prove it on real data. Key job responsibilities Own the scientific strategy for how organizational knowledge is represented, kept current, and retrieved: extraction, entity resolution, deduplication, graph structure, and retrieval that unifies graph, semantic, keyword, and temporal search. Advance temporal reasoning. Knowledge changes: facts are revised, decisions are reversed, priorities move. Representing what superseded what and when, and preserving the provenance to distinguish confirmed information from inferred information, is among the hardest open problems in this space. Define the science of proactive behavior. When is it right for an AI system to interrupt a human? These are precision-critical problems where a false positive costs far more than a miss, and where the right threshold varies by team and by individual. Lead our measurement science. Build evaluation for completeness and correctness across a multi-component agentic system, converging on a small number of trustworthy primary metrics rather than a sprawl of component scores. Judge honestly when an offline gain is real and when it is an artifact of a sparse dataset. Build the data that doesn't exist. The most valuable phenomena in this domain are also the rarest, which makes naturally occurring examples too scarce to learn from. Design synthetic and simulated data pipelines that generate controlled, realistic scenarios so these capabilities can be developed and tested at all. Own the learning loop. Turn human interaction into usable training signal, and set the direction for how the system improves from explicit feedback in the near term and from passive observation over the longer term. Make the efficiency calls. Decide where frontier models are required and where a smaller domain-tuned model is sufficient, and build the cost and capacity measurement that makes it a data-driven decision rather than an opinion. Raise the bar across the team. Mentor scientists, review designs, publish where the work merits it, and represent the science externally to customers and to the research community. A day in the life You might spend the morning in a design review arguing that a proposed approach won't survive contact with real data, the afternoon writing a prototype yourself to demonstrate the alternative, and the end of the day convincing an engineer that the capability is worth a sprint. Our sequencing is deliberate: try the idea on intuition, validate it on real data by inspection, then measure it, then operationalize it. Scientists here are expected to identify a problem, justify it, recruit others to it, and drive it into production, across whatever parts of the system that requires. Ownership follows the problem, not the org chart. About the team We are a combined science, product, and engineering team building one product together. Scientists own capabilities end to end rather than individual components, because these problems don't decompose cleanly: a single improvement typically touches extraction, storage, and retrieval at once. We invest in the tooling that makes that practical: local full-stack environments and sandboxed realistic data, so a scientist can go from idea to result in seconds rather than waiting on a deployment or on engineering support. The work is grounded in real usage rather than benchmarks alone, which is a rare combination for science this early: real users, real data, real feedback, and a genuinely unsolved research agenda.
2027 Applied Science Internship - Automated Reasoning - United States, PhD Student Science Recruiting, Automated Reasoning
US, WA, Seattle
This role sits within Amazon's Automated Reasoning and Formal Verification research horizon. Shape the Future of Cloud Computing. Are you a graduate student passionate about Automated Reasoning and its real-world applications? Join our team of innovators and embark on a journey to revolutionize cloud computing through innovative automated reasoning techniques. Our tools are called billions of times daily, powering the backbone of Amazon's products and services. We are changing the way computer systems are developed and operated, raising the bar for security, durability, availability, and quality. Applied Scientists in Automated Reasoning develop and apply formal methods, automated reasoning techniques, and neurosymbolic approaches to ensure the security, reliability, and correctness of Amazon and AWS services and customer applications. Application areas span cloud infrastructure verification, cryptographic assurance, AI safety, and formal guarantees for generative AI systems. Methods range from interactive theorem proving and constraint solving to neuro-inspired proof search. As an Applied Science Intern, you will have the opportunity to work alongside our scientists and contribute to projects. From distributed proof search and SAT/SMT solvers to program analysis, synthesis, and verification, you will tackle complex challenges at the intersection of theory and practice. Amazon has positions available for Automated Reasoning Applied Science Internships in, but not limited to, Arlington, VA; Boston, MA; New York, NY; Portland, OR; Santa Clara, CA; Seattle, WA; Austin, TX; Cambridge, UK. Key job responsibilities We are particularly interested in candidates with expertise in: Theorem Proving, Boolean Satisfiability Solvers, Bounded Model Checking, Deductive Verification, Programming/Scripting Languages, Abstract Interpretation, Automated Reasoning, Static/Program Analysis, Program Synthesis. Contribute to the design and implementation of algorithms and formal methods for automated reasoning, including constraint solving, model checking, static analysis, theorem proving, and program synthesis, within a guided research framework. Explore and apply generative AI and machine learning techniques to enhance automated reasoning, including learning-based heuristics for search, neural approaches to symbolic reasoning, and methods for verifying the correctness of AI-generated code. Contribute to automated reasoning techniques for generative AI and agentic coding systems, including methods that apply formal guarantees to large language model outputs. Contribute to the scientific community through publications at peer-reviewed conferences and journals. Leverage AI-powered tools where applicable to accelerate research, experimentation, and prototyping. Critically review and validate outputs from AI tools and automated systems. The ideal intern must have the ability to communicate research findings clearly to diverse audiences.
Sr. Applied Scientist, Pricing Science
US, WA, Seattle
We are looking for a talented, organized, and customer-focused applied researcher to join our Pricing Optimization science group, with a charter to measure, refine, and launch customer-obsessed improvements to our algorithmic pricing and promotion models across all products listed on Amazon. This role requires an individual with exceptional machine learning modeling and architecture expertise — particularly in deep learning, neural networks, and transformer-based architectures applied to price prediction and forecasting problems. Equally important is deep expertise in causal machine learning — including causal inference, treatment-effect estimation, and experimentation methods (e.g., uplift modeling, double/debiased machine learning, instrumental variables, and A/B and quasi-experimental design) — to isolate the true impact of pricing and promotion decisions on customer behavior and business outcomes. The ideal candidate brings a strong foundation in applied statistics and probabilistic modeling, excellent cross-functional collaboration skills, business acumen, and an entrepreneurial spirit. We are looking for an experienced innovator who is a self-starter, comfortable with ambiguity, demonstrates strong attention to detail, and has the ability to work in a fast-paced and ever-changing environment. Key job responsibilities See the big picture. Understand and influence the long-term vision for Amazon's science-based competitive, perception-preserving pricing techniques. Develop and advance price prediction models leveraging deep learning frameworks, transformer architectures, and advanced statistical methods to drive pricing accuracy at scale. Build strong collaborations. Partner with product, engineering, and science teams within Pricing & Promotions to deploy machine learning price estimation and error correction solutions at Amazon scale. Design and implement neural network-based architectures — including sequence models and transformers — for large-scale price prediction and optimization. Stay informed. Establish mechanisms to stay up to date on the latest scientific advancements in deep learning, transformer architectures, applied statistics, neural network design, probabilistic forecasting, and multi-objective optimization techniques. Identify opportunities to apply them to relevant Pricing & Promotions business problems. Keep innovating for our customers. Foster an environment that promotes rapid experimentation, continuous learning, and incremental value delivery. Leverage statistical rigor and modern deep learning approaches to validate hypotheses and drive measurable pricing improvements. Successfully execute & deliver. Apply your exceptional technical machine learning expertise — including deep neural networks, attention-based models, and applied statistical analysis — to incrementally move the needle on some of our hardest pricing problems. A day in the life We are hiring a Sr. Applied Scientist to drive our pricing optimization initiatives. We drive cross-domain and cross-system improvements through: * shape and extend our RL optimization platform - a pricing centric tool that automates the optimization of various system parameters and price inputs. * Error detection and price quality guardrails at scale. * Identifying opportunities to optimally price across systems and contexts (marketplaces, request types, event periods) Price is a highly relevant input into Stores architectures; this role creates the opportunity to drive extremely large impact (measured in Bs not Ms), but demands careful thought and clear communication. About the team The Pricing Optimization science group builds and refines Amazon's algorithmic pricing and promotion models at scale. Our team combines expertise in deep learning, transformer architectures, applied statistics, and probabilistic forecasting to develop price prediction systems that directly impact the customer experience. The team also brings hands-on experience with causal modeling and inference — including uplift modeling and treatment effect estimation — to rigorously measure the impact of pricing decisions on customer behavior and business outcomes. We partner closely with product, engineering, and business teams to take solutions from research through production deployment.
Applied Scientist (Speech AI), Content Reasoning, Enrichment and Localization
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 subscriptions such as Apple TV+, HBO Max, Peacock, 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! We are looking for passionate, hard-working, and talented individuals to help us push the envelope of content localization. We are seeking scientists with experience in audio processing, speech/voice AI and machine learning. We work on a broad array of research areas and applications, including but not limited to multimodal machine translation, speech synthesis, speech analysis, and asset quality assessment. Candidates should be prepared to help drive innovation in one or more areas of machine learning, audio processing, and natural language understanding. If you have experience with speech synthesis and foundational models, then that's a huge plus! Key job responsibilities As an Applied Scientist, you should be a strong communicator, able to describe scientifically rigorous work to business stakeholders of varying levels of technical sophistication. You will closely partner with the solution development teams, and should be intensely curious about how the research is moving the needle for business. Strong inter-personal and mentoring skills to develop applied science talent in the team is another important requirement. - Lead research and development of speech and audio generation technology and end-to-end speech-to-speech architecture - Develop audio processing solutions for production environments, including source separation, enhancement, and mixing - Define the research roadmap for your area, identify high-impact problems, and communicate technical direction to senior leadership - Publish research, contribute to the broader scientific community, and bring external advances into production systems A day in the life You might start your morning reviewing experimental results and refining a model architecture before syncing with your engineering partners on integration plans. After lunch, you could be whiteboarding a new approach to a problem your team recently identified, then writing up findings for an internal science review. You will regularly present your work to peers and stakeholders, participate in code and design reviews, and explore emerging research that could unlock new possibilities for your team. About the team Our team is driven by a shared commitment to applying science in ways that create meaningful impact for customers. We value rigorous research, collaborative problem-solving, and a willingness to experiment with new ideas. You will work alongside talented scientists and engineers in an inclusive environment where your contributions shape the direction of our work. We are focused on building solutions that matter at scale, and we are looking for teammates who are energized by that challenge.
Senior Applied Scientist, Fauna
US, NY, New York
We are seeking an Applied Scientist to lead the development of evaluation frameworks and data collection protocols for robotic capabilities. In this role, you will focus on designing how we measure, stress-test, and improve robot behavior across a wide range of real-world tasks. Your work will play a critical role in shaping how policies are validated and how high-quality datasets are generated to accelerate system performance. You will operate at the intersection of robotics, machine learning, and human-in-the-loop systems, building the infrastructure and methodologies that connect teleoperation, evaluation, and learning. This includes developing evaluation policies, defining task structures, and contributing to operator-facing interfaces that enable scalable and reliable data collection. The ideal candidate is highly experimental, systems-oriented, and comfortable working across software, robotics, and data pipelines, with a strong focus on turning ambiguous capability goals into measurable and actionable evaluation systems. Key job responsibilities - Design and implement evaluation frameworks to measure robot capabilities across structured tasks, edge cases, and real-world scenarios - Develop task definitions, success criteria, and benchmarking methodologies that enable consistent and reproducible evaluation of policies - Create and refine data collection protocols that generate high-quality, task-relevant datasets aligned with model development needs - Build and iterate on teleoperation workflows and operator interfaces to support efficient, reliable, and scalable data collection - Analyze evaluation results and collected data to identify performance gaps, failure modes, and opportunities for targeted data collection - Collaborate with engineering teams to integrate evaluation tooling, logging systems, and data pipelines into the broader robotics stack - Stay current with advances in robotics, evaluation methodologies, and human-in-the-loop learning to continuously improve internal approaches - Lead technical projects from conception through production deployment - Mentor junior scientists and engineers About the team Fauna Robotics, an Amazon company, is building capable, safe, and genuinely delightful robots for everyday life. Our goal is simple: make robots people actually want to live and interact with in everyday human spaces. We believe that future won’t arrive until building for robotics becomes far more accessible. Today, too much effort is spent reinventing the fundamentals. We’re changing that by developing tightly integrated hardware and software systems that make it faster, safer, and more intuitive to create real-world robotic products. Our work spans the full stack: mechanical design, control systems, dynamic modeling, and intelligent software. The focus is not just functionality, but experience. We’re building robots that feel responsive, expressive, and genuinely useful. At Fauna, you’ll work at the frontier of this space, helping define how robots move, manipulate, and interact with people in natural environments. It’s an opportunity to solve hard problems across hardware and software with a team focused on making robotics accessible and joyful to build. If you care about making robotics real for everyone and building systems that are as delightful as they are capable, we’re interested in hearing from you.
See more jobs
View from space of a connected network around planet Earth representing the Internet of Things.
Get more from Amazon Science
Subscribe to our newsletter

Amazon.com | Conditions of Use | Privacy | © 1996-2026 Amazon.com, Inc. or its affiliates

Social