Five ways the ABACUS label advances nature-based carbon removal

From more-accurate measurement of carbon dioxide removal to greater diversity in restoration design, the ABACUS label’s requirements help advance the integrity of restoration projects in the voluntary carbon market.

Amazon cofounded the Climate Pledge in 2019 to commit to reaching net-zero carbon by 2040. The first priority of the pledge is to implement decarbonization strategies — in line with the Paris Agreement — through operational changes such as improving efficiency, driving forward scalable carbon-free energy sources, reducing waste, and innovating materials.

However, alongside real business change that directly reduces greenhouse gas emissions, there is also need for large-scale investments in climate change mitigation outside of our value chain (what we call carbon neutralization). At Amazon, we do this through both nature-based solutions and technological carbon dioxide removal.

Nature-based carbon removal harnesses the power of photosynthesis to sequester carbon in natural and managed ecosystems. This means altering land management in alignment with nature through native reforestation, agroforestry, and other forms of high-quality restoration. These activities alone have the potential to remove 2–4 billion tons of carbon per year; that’s almost half of the estimated 5–10 billion tons per year that experts estimate is likely needed through the end of the century in order to keep our global temperatures at safe levels.

While the voluntary carbon market has the potential to bring billions of dollars of finance to restoration projects, less than 3% of credits issued to date come from nature-based carbon removal. This is due to the voluntary carbon market’s prices’ falling below the costs of high-quality nature-based restoration.

That’s where ABACUS comes in. ABACUS is a set of principles and requirements, codified within Verra’s Verified Carbon Standard, that helps advance the integrity of restoration projects within the voluntary carbon market. ABACUS was developed by a working group of expert practitioners, conservation professionals, and scientists — including Amazon’s own carbon neutralization scientists — in an effort to raise the quality bar for agroforestry and native-restoration projects. The ABACUS label has already begun to raise the quality bar for leading buyers.

Below are five big ideas within ABACUS that help raise the bar on scientific rigor and transparency.

  1. Dynamic baseline to measure additionality

    Historically, restoration carbon projects assume that whatever land use was occurring before a project takes place — pasture or agriculture, for example — would have continued unaltered without the project intervention. This assumption ignores the myriad ecological, economic, and policy dynamics that could affect carbon removal without assistance from the voluntary carbon market.

    Related content
    Investing in 500+ solar and wind projects, bringing carbon-free energy to dirty grids, and buying Renewable Energy Certificates all played a role.

    In addition to demonstrating that a project would not be viable without carbon credit finance, ABACUS requires a treatment-control approach to measuring additionality, or the carbon removal resulting from the project above and beyond what would have occurred otherwise. This means matching the project “treatment” area — based on historical, satellite-based proxies for biomass — to a population of “control” plots that are followed through time. Each of these controls represents a potential alternate reality for the project in the absence of restoration.

    If the control plots regain forest carbon at pace with the project, this indicates that the project may have regained forest carbon on its own, without the intervention. If the control plots remain low-carbon, degraded land, we can be more confident that the project’s climate impacts are additional. By treating additionality as dynamic instead of static, we’re able to obtain a more data-driven estimation of the true impact of restoration.

  2. Carbon projects as engines for agricultural production

    Carbon removal cannot come at the expense of food production; in fact, these challenges are inextricably linked. Under some projections, agricultural production will need to double by 2050, even as the least productive pasture and croplands are restored to forest cover. Sustainably intensifying agriculture to increase food production, while sparing land for carbon removal — or, better, integrating carbon removal within productive agricultural systems — is critical to reconciling these needs.

    Drone footage of a mature cocoa, coconut, and mahogany agroforestry system, adjacent to a degraded pasture in southeast Pará, Brazil.
    ABACUS seeks to restore degraded pasturelands to diverse agroforestry systems like this one. (Drone footage courtesy of Eric Plançon)

    But the voluntary carbon market is not equipped to tackle this challenge. Carbon removal projects that displace agricultural production often result in indirect land use change and associated emissions, as agricultural markets replace lost production to serve growing demand (“leakage”).

    These crop- and region-specific leakage effects are difficult to quantify reliably. Conventional leakage methodologies impose standardized deductions based on default carbon leakage rates when agricultural production is displaced. This creates a persistent source of uncertainty and risk of over-crediting, and the approach misses an opportunity to build synergies between restoration and agricultural production.

    Related content
    From investing in new carbon-free energy projects to advocating for grid modernization and collaborating with key stakeholders around the world, Amazon is working toward a cleaner energy future.

    ABACUS instead takes a “food-forward” approach to leakage accounting. Rather than using an imprecise default value to quantify leakage effects, ABACUS requires projects to eliminate leakage by maintaining or enhancing agricultural production in the project areas and surrounding landscapes. By recognizing the land-sparing effect of enhancing production of different types of commodities, ABACUS encourages projects to co-optimize for carbon and agricultural production and avoids locking regions into specific agricultural products. The working group is engaging partners to create commodity-specific leakage metrics based on land-carbon “opportunity costs” to estimate, and mitigate, the impacts of leakage.

  3. Abbreviated crediting periods for durability assurance

    Carbon stored in ecosystems can be highly durable, but it faces persistent, long-term climate risks such as fire, drought, and land use change, which must be responsibly managed. Nature-based carbon removal should seek “effective permanence” — an actual net greenhouse gas benefit to the atmosphere that is equal to, or greater than, the net benefit represented by the credits. In addition, the removal should ensure that this balance can be maintained indefinitely.

    On the other hand, agroforestry and restoration projects can catalyze shifts to land use systems that durably enhance carbon storage even beyond what is credited. This can happen through spillover effects, continued carbon removal after the crediting period, and biophysical cooling feedbacks, among other factors. ABACUS includes several methods that improve the likelihood that nature-based carbon remains durably stored — for example, requiring projects to plant ecologically appropriate restoration systems and to create public plans for the longevity of project activities even after the support of carbon revenues.

    Related content
    Amazon teams up with RTI International, Schlumberger, and International Paper on a project selected by the US Department of Energy to scale carbon capture and storage for the pulp and paper industry.

    One of ABACUS’s key innovations is to limit the crediting period in an effort to maximize uncredited removals. The ABACUS working group found that revenues from credits generated beyond year 30 are mostly immaterial to investment decisions today, due to their heavy discounts. By shortening the crediting period to 40 years maximum — as opposed to as much as 100 years under some voluntary carbon market standards — ABACUS will create a source of uncredited carbon removal that can serve as an additional buffer against future reversals.

    Additionally, ABACUS proposes that projects will be required to allocate a portion of carbon credits issued late in the crediting period (i.e., years 31–40) to a “long-term permanence mechanism” such as an enhanced buffer pool or insurance product. Achieving increased confidence in the effective permanence of nature-based carbon credits may require stringing together removals or replacing a moderate-durability credit with a high-durability credit, if and when previously credited removals are reversed. Economically, such a construct is currently likely to be cost effective compared to today’s high-durability carbon dioxide removal.

  4. Going beyond commercial monoculture plantations

    Forest plantations already cover nearly 300 million hectares globally — roughly equivalent to the entire area of India. That figure has more than doubled in the last 30 years, without a robust voluntary carbon market, and it is projected to continue growing to provide timber, pulpwood, firewood, and charcoal to increasing populations and a growing economy.

    Brazil_Drone.png
    Orthorectified mosaic capturing a range of land management types on a typical farm in the Amazon basin, Brazil. We can see the contrast between low-carbon-density pasture (left) and diverse agroforestry (center), which combines shade-tolerant commodity production with native, carbon-rich hardwood trees. ABACUS is designed to support native restoration and agroforestry interventions on formerly forested, degraded land.
    Photos captured and combined by ICRAF-Brazil on behalf of the Agroforestry Accelerator.

    As a first step, ABACUS prohibits most monocultures and requires project developers to use observed or modeled data to demonstrate that planted systems are ecologically appropriate for the landscape. This approach avoids projects seeking to reforest with systems that aren’t suitable for the location’s native biomass potential — a function of climate, soil type, water availability, and elevation, among other things. Credit buyers are encouraged to send demand signals that further encourage biodiverse, ecologically sound, and socially beneficial restoration.

  5. Transparency to foster competition on quality

    For some aspects of restoration, it’s challenging to prescribe universally applicable requirements without stifling innovation and local knowledge: every restored ecosystem is unique in its own way. ABACUS introduces multiple requirements for added transparency that will allow buyers, investors, and the public to better assess for themselves the effectiveness of project designs and measurement.

    Related content
    Amazon advocates for updating carbon accounting to measure where renewable-energy projects will have the greatest impact.

    For example, ABACUS projects will need to publish their in-situ inventory measurements, systematically justify their use of allometric or other scaling models, and report on design approaches to avoid measurement or sampling bias. Instead of once every five years or so, ABACUS requires projects to annually map disturbances, to ensure that carbon credited and subsequently reversed is immediately identified. With enhanced transparency, the ABACUS working group hopes to incentivize project developers to compete on quality.

  6. ABACUS doesn’t solve all of the challenges of quantifying the complete climate impact of nature-based carbon removal, and it is no replacement for the stakeholder engagement necessary to ensure genuine socio-economic benefits on the ground. Many important improvements remain for future versions of the label’s principles and requirements. As we learn, the ABACUS working group will continue to enhance the scientific rigor of and public confidence in ecosystem restoration, catalyzing rural restoration economies and livelihoods and — if we succeed — helping to enable billions of tons of ecosystem carbon removal across the world.

Research areas

Related content

US, NY, New York
We are seeking a Human-Robot Interaction (HRI) Research Scientist to develop cutting-edge interactions that make robots feel alive, personal, and fun. In this role, you will focus on verbal and non-verbal conversational systems, social dynamics, memory, and long-term relationship formation between robots, their environments, and the people they interact with. Your contributions will be essential in advancing robotics by enabling expressive, socially intelligent, and trustworthy interactions between robots and humans.
US, MA, North Reading
Amazon Robotics is transforming warehouse automation through edge AI and machine learning applied to real-world robotics challenges. We're seeking a Research Scientist to advance our mobile manipulation capabilities by developing novel learning-based approaches that enable robots to navigate and manipulate objects in dynamic fulfillment environments. This role offers the opportunity to conduct original research and translate state-of-the-art findings into production systems operating at Amazon's unprecedented scale. Key job responsibilities Research and Algorithm Development: Formulate novel research problems in robot learning and manipulation, design new model architectures, validate hypotheses through rigorous experimentation, and advance the state of the art in learning-based robotics. Data Strategy and Pipeline Design: Define data requirements for research initiatives, design scalable collection and curation strategies, establish governance and provenance standards, and build reusable pipelines ensuring data quality and reproducibility. Experimentation and Scientific Validation: Design and execute experiments in simulation and real-world embodiments, develop evaluation methodologies and benchmarks, perform ablation studies and statistical analyses, and iterate systematically to advance model performance. Prototyping and Research Infrastructure: Develop clean, well-documented research codebases, build experimentation frameworks and evaluation tooling, contribute to shared training infrastructure, and implement interfaces for broader robotics integration. Scientific Leadership and Publication: Drive an independent research agenda aligned with team objectives, publish at top-tier venues (e.g., RSS, CoRL, ICRA, NeurIPS), identify research gaps through literature reviews, and present findings via technical reports and talks. Cross-Functional Collaboration: Partner with scientists, engineers, and leaders across teams to translate research into deployable solutions, mentor junior researchers, contribute to the team's scientific culture, and support integration with robotics hardware teams. A day in the life 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! About the team Are you inspired by invention? Is problem solving through teamwork in your DNA? Do you like the idea of seeing how your work impacts the bigger picture? Answer yes to any of these and you’ll fit right in here at Amazon Robotics. We are a smart, collaborative team of enthusiastic doers that work passionately to apply innovative advances in robotics and software to solve real-world challenges that will transform our customers’ experiences in ways we can’t even image yet. We invent new improvements every day. We are Amazon Robotics and we will give you the tools and support you need to invent with us in ways that are rewarding, fulfilling and fun!
US, NY, New York
Want to work on building a Amazon Ads billion dollar business, innovate on a new product, and have a positive impact on millions of views while working with industry-leading technologies? We're growing a team to support the Sponsored Ads business that powers the advertising experience for millions of viewers and advertisers daily. Amazon is investing heavily in building a world-class advertising business and developing a collection of self-service performance advertising products that drive discovery and sales. We deliver billions of ad impressions and millions of clicks daily and are constantly challenging ourselves to create world-class products and an unparalleled shopping experience for our hundreds of millions of customers worldwide. Key job responsibilities We are building the next-gen smart ads campaign. At its core is an Intelligence Flywheel — an architecture where every component's output is designed to train models that improve every other component. The Model Layer that is meant to power every capability currently has no dedicated science ownership. As the Senior Applied Scientist on this team, you own the science that makes the flywheel turn. You will turn static, threshold-based logic into self-improving, closed-loop intelligence, and define the decision policies that let the system act autonomously with advertiser trust. Concretely, you will: * Build predictive issue-detection models that identify under-delivery, over-delivery, and performance degradation from campaign signals before they materially impact advertisers. * Design the intervention-selection policy — which autonomous action to take — framed as a contextual bandit: choose, observe, update. Engineers implement action execution; you define the policy that selects actions. * Establish causal attribution for autonomous optimization, separating the effect of our interventions from organic performance change, so improvements can be attributed and advertiser trust in autonomy can be earned. * Integrate and adapt cross-model signals. Combine creative-quality, product-relevance, and budget signals into a unified advertiser-intelligence picture, and adapt general-purpose partner models to our product reality via fine-tuning, re-ranking, or thin adaptation layers. * Close recommendation and grading feedback loops — define reward schemas for accept/reject and performance-vs-baseline signals, correlate creative-quality scores with real campaign outcomes, and feed empirical findings back to both our product and partner science teams. You will work backwards from ambiguous business problems, set the science roadmap for the Model Layer, and partner closely with the team's software engineers — who own the services, pipelines, and execution infrastructure — so that model artifacts you produce are deployed and served in production. This is a high-leverage, high-autonomy role: your outputs are consumed by multiple engineering workstreams at once, and you set the abstractions the team builds on. About the team We are focused on goal-oriented, AI powered workflows that help advertisers achieve their marketing objectives. We collect campaign goals, surface relevant data at key decision points, and provide reporting that validates decision-making. Our product suite guides advertisers in building campaigns with optimal targeting, creative formats, inventory, and bid models that are highly likely to hit their goals — reducing the need for manual intervention.
US, WA, Seattle
At Amazon Selection and Catalog Systems (ASCS), our mission is to power the online buying experience for customers worldwide so they can find, discover, and buy any product they want. We innovate on behalf of our customers to ensure uniqueness and consistency of product identity and to infer relationships between products in Amazon Catalog to drive the selection gateway for the search and browse experiences on the website. We're solving a fundamental AI challenge: establishing product relevant information at unprecedented scale with Frontier Models and Agents. The scale is staggering: billions of products, petabytes of multimodal data, millions of sellers, dozens of languages, and infinite product diversity ranging from electronics to groceries to digital content. The research challenges are immense. GenAI and VLMs hold transformative promise for catalog understanding, but we operate where traditional methods fail: ambiguous problem spaces, incomplete and noisy data, inherent uncertainty, reasoning across both images and textual data, and explaining decisions at scale. Enriching product information requires sophisticated models that reason across text, images, and structured data, all while maintaining accuracy and trust for high-stakes business decisions affecting millions of customers daily. Amazon's Catalog System Services Science team is looking for an innovative and customer-focused applied scientist to help us make the world's best product catalog even better. In this role, you will partner with technology and business leaders to build new state-of-the-art algorithms, models, and services. You will pioneer advanced GenAI solutions that power next-generation agentic shopping experiences, working in a collaborative environment where you can experiment with massive data from the world's largest product catalog, tackle problems at the frontier of AI research, rapidly implement and deploy your algorithmic ideas at scale, across millions of customers. Key job responsibilities - Formulate novel research problems at the intersection of GenAI, multimodal learning, and large-scale information retrieval. In essence, translating ambiguous business challenges into tractable scientific frameworks - Design and implement leading models leveraging frontier models, and agentic architectures to enrich catalog information at billion-product scale - Pioneer explainable AI methodologies that balance model performance with scalability requirements for production systems impacting millions of daily customer decisions - Own end-to-end ML pipelines from research ideation to production deployment, processing petabytes of multimodal data with rigorous evaluation frameworks - Represent the team in the broader science community - publishing findings, delivering tech talks, and staying at the forefront of GenAI, VLM, and agentic system research
US, WA, Seattle
What happens when you give AI the ability to remember? Not cached responses — real structured memory that compounds over time and transfers across contexts. We're building the science behind this, and we need researchers who want to own the problem end-to-end. This is a founding role on a new team. You won't inherit models or maintain someone else's pipeline. You'll define the research direction, run experiments at scale, and ship what works directly to production. Key job responsibilities As an Applied Scientist in our team, you will be responsible for the research, design, and development of new AI technologies for knowledge acquisition and retrieval. You will adopt or invent new machine learning and analytical techniques in the realm of information retrieval, knowledge representation, and large language models. Specific responsibilities include: 1. Design and implement novel approaches to knowledge extraction from heterogeneous, unstructured data sources at organizational scale. 2. Build retrieval systems that match intent to relevant knowledge across domains — solving the "right memory at the right time" problem. 3. Own the quality of memory generation: what to capture, how to structure it, when to surface it, and when to let it decay. 4. Run large-scale experiments using Amazon's compute infrastructure and massive real-world datasets. 5. Develop evaluation frameworks for a system where "quality" means something new — right knowledge, right context, right confidence level. 6. Collaborate with engineers to move from research prototype to production system in weeks, not quarters. 7. Invent new approaches to temporal knowledge management — how memories age, conflict, and compound over time. 8. Publish and patent novel approaches to knowledge acquisition and retrieval at top-tier venues. A day in the life You will solve real-world problems by getting and analyzing large amounts of data, generate insights and opportunities, execute experiments, and develop statistical and ML models. The team is driven by business needs, which requires collaboration with other Scientists, Engineers, and Product Managers across the organization. You get to influence stakeholders with clear communication skills. You innovate on behalf of the customer and strategically build features. You will mentor junior members and help them grow. About the team We're a new team within Personalization, focused on a different kind of recommendation: not "what product should this customer see" but "what knowledge should this AI use right now." Same scale, same rigor, entirely new problem space. The science is at the intersection of information retrieval, knowledge representation, and LLM reasoning — and the right approach hasn't been established yet. The team values innovation and offers a safe place to try, fail, and learn while fostering a culture of continuous improvement. Everyone is a leader and owner for everything we do as a team. We offer creative space with an entrepreneurial work environment focusing on customer obsession.
IN, HR, Gurugram
Building large-scale forecasting and optimization systems that power Amazon’s global transportation network and directly impact customer experience and cost. Key job responsibilities 1. Guide model and system design across a range of techniques, including tree-based models, deep learning (LSTMs, transformers), LLMs, and reinforcement learning. 2. Ensure models are production-ready, scalable, and robust through close partnership with stakeholders. 3. Partner with Product, Operations, and Engineering leaders to enable proactive decision-making and corrective actions. 4 Own end-to-end business metrics, directly influencing customer experience, cost optimization, and network reliability. 5. Help contribute to the broader ML community through publications, conference submissions, and internal knowledge sharing.
US, CA, Sunnyvale
We are looking for a Senior Applied Scientist to help drive the research and development of real-time multimodal conversational AI. You will contribute across two focus areas: advancing foundation models for speech and audio, and building the post-training systems (reward modeling, reinforcement learning) that shape natural, human-like conversational behavior. You will own a significant research area and contribute across the full model lifecycle — from pre-training and architecture design through post-training alignment and real-time deployment. You will work at the frontier of what’s possible in conversational AI, with the compute, data, and runway to pursue problems that few teams in the world have the resources to tackle. As a Senior Scientist, you will drive the technical execution of your research area, contribute to the team’s roadmap, and work closely with inference engineers to ensure your models are designed for real-time production deployment. Key job responsibilities What You’ll Do Foundation Model Scaling - Help build and train large-scale multimodal foundation models for real-time speech and audio generation, from architecture design through production-scale training - Advance the scaling and efficiency of conversational models, including the relationship between data, model size, and real-time performance - Design model architectures informed by hardware constraints and inference requirements, working with inference engineers to ensure models are servable from inception - Develop training methodologies for multimodal models that jointly process and generate speech, language, and audio in real-time streaming contexts - Contribute to the state of the art on efficient architectures and training methods for conversational AI at scale Post-Training & Reinforcement Learning - Design and build reward models and reward functions for speech systems — capturing naturalness, fluency, conversational quality, and real-time responsiveness - Develop and apply reinforcement learning methods to shape conversational behavior — teaching models natural timing, responsiveness, and fluid interaction - Build parts of the post-training pipeline from SFT through RL alignment, optimized for real-time multimodal outputs rather than text-only generation - Design evaluation frameworks that capture the quality dimensions unique to real-time conversation (latency sensitivity, audio quality, prosody, interaction naturalness) Real-Time Perception & Generation - Advance the team’s capabilities in real-time perception — the ability of the model to process incoming audio/speech while simultaneously generating responses - Develop techniques for natural interactive systems where the model handles concurrent input and output with human-like timing - Work at the intersection of model architecture and production constraints to ensure multimodal capabilities function within hard real-time latency budgets
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!
US, NY, New York
Are you excited about applying machine learning and statistical modeling to real-world systems that serve millions of customers? Amazon Connect is a cloud-based contact center service that helps businesses deliver personal, efficient customer experiences. Our team of scientists and engineers builds the AI and ML capabilities that power contact center operations and optimization. We are looking for a Senior Applied Scientist to tackle scientifically complex challenges in areas such as stochastic modeling, queueing theory, anomaly detection, and optimization. In this role, you will design and deploy novel ML models and algorithms that directly improve how businesses interact with their customers. You will work at the intersection of research and production, turning ambiguous problems into scalable solutions that shape the future of cloud-based customer service. Key job responsibilities - Design and deploy novel machine learning models and algorithms to solve complex problems in contact center operations, including forecasting, routing optimization, and anomaly detection. - Lead the scientific agenda for your team by identifying new research opportunities, proposing initiatives, and driving them from concept through production deployment. - Collaborate with engineering teams to architect and implement scalable ML systems, personally contributing significant portions of the critical scientific components. - Mentor fellow scientists and engineers through code reviews, design discussions, and scientific guidance, raising the overall technical bar of the team. - Evaluate and advance the team's ML methodology by benchmarking against current academic and industry research, and by publishing findings internally and externally when appropriate. A day in the life You might start your morning reviewing experiment results from a new forecasting model, then join a design session with engineers to discuss how to integrate it into the production pipeline. After lunch, you could be whiteboarding a novel approach to a queueing optimization problem with a fellow scientist, followed by a code review for a teammate. You will regularly present your research findings to stakeholders across the organization and contribute to the team's publication efforts. About the team Our team within Amazon Connect focuses on building intelligent, ML-driven capabilities that help businesses run their contact centers more effectively. We work closely with product, engineering, and science partners to turn research ideas into features that customers rely on every day. We value curiosity, collaboration, and scientific rigor, and we are investing in new AI capabilities that will continue to transform the customer service industry. If you want to see your research make a tangible impact at scale, this is the place to do it.
ES, B, Barcelona
How does Amazon decide which fulfillment center ships your order, which truck carries it, and how to keep promises across hundreds of millions of packages daily? How does it decide how many trucks and how much labor are required to ship orders across the network? SCOT Fulfillment Optimization (FO) owns the optimization and forecasting science behind these decisions. We are seeking Applied Scientists to join the FO Science & Tech team in Barcelona (alternatively: Luxembourg or London) with a strong academic background in optimization, machine learning, and/or time-series forecasting. • You will design and build state-of-the-art machine learning and optimization models that power Amazon's fulfillment decisions at an unprecedented scale across two core scientific pillars: • Large-Scale Optimization and Planning: Designing planning systems for order assignment and resource utilization, while balancing multi-objective cost-speed tradeoffs to enable controllers to steer millions of shipments per hour optimally. • Demand Forecasting & Predictive ML: Developing time-series forecasts for customer demand, incorporating contextual information (weather, sales, order properties), and modeling uncertainty for core planning systems. Basic qualifications • PhD in Operations Research, Applied Mathematics, Computer Science, or related field (or equivalent experience) • Strong programming skills (Python preferred; experience with optimization solvers a plus) • Research experience in one or more: • Large-scale mathematical programming (LP, MIP, decomposition methods) • Combinatorial optimization (assignment, scheduling, network flows) • Multi-objective optimization and control • Large-scale time-series forecasting (GenAI models, probabilistic forecasting, uncertainty quantification) • Causal inference (spatiotemporal causal modeling, offline policy evaluation) Preferred qualifications • Experience building optimization systems that run in production at scale • Being comfortable with ambiguity and fast iteration cycles • Publications in relevant venues Key job responsibilities Design and implement optimization and forecasting models for large-scale fulfillment problems, from order assignment to network flow control. Build research prototypes end-to-end: from problem formulation through scalable implementation to production validation. Analyse complex tradeoffs (cost, speed, capacity, accuracy) and translate findings into actionable recommendations for leadership and operations teams. Collaborate with engineers to bring science solutions into production systems serving millions of customer orders daily. A day in the life You formulate an optimization or forecasting problem on a whiteboard with teammates, then prototype it in Python with real data by the afternoon. You run experiments against production-scale datasets, iterate on the model, and present results to stakeholders who will use them to make network decisions next week. Some days you dive deep into solver performance; other days you're explaining a Pareto frontier to an operations leader. You collaborate with large engineering and product teams to bring your solutions into systems serving millions of customers. Alongside fast-turnaround prototypes, you own long-term research bets, the kind that reshape how Amazon's fulfillment network operates at scale. Your work goes live. About the team SCOT Fulfillment Optimization Science & Tech (FO SnT) is the applied research team behind Amazon's fulfillment decision-making systems. We decide how orders get assigned to warehouses, how capacity is allocated across the network, and how cost and speed tradeoffs are managed in real time, at global scale. Our models influence billions of euros in annual operational spend. They protect sites from overload during peak, reduce transportation costs and CO2 emissions, and ensure customers receive their packages when promised. Leadership relies on our science to make investment decisions worth hundreds of millions. We are practitioners of large-scale optimization: MIP formulations, decomposition methods, approximation algorithms, and parallelisation. We use machine learning where it sharpens our decisions, including forecasting, learned heuristics, and multi-armed bandits. We pick the right tool for the problem, not the fashionable one. You will work alongside Senior and Principal scientists, and collaborate with Amazon Scholars and academic partners who bring frontier research into our applied problems. We code our prototypes to be production-ready and collaborate with large engineering teams to ship systems, not papers. Above all, we have fun solving hard real-world problems at real-world speed, failing, learning, and shipping along the way.