Amazon Science

AI agents are rapidly scaling: longer sessions, deeper reasoning chains, more autonomy. But the Transformer architecture underneath them still pays a memory tax that grows with the number of tokens it processes. As agents become more capable, that cost scales with them. Now, there’s a proliferation of hybrid architectures that use some combination of methods to achieve a better balance of memory, speed, and quality.

Amazon and the Georgia Institute of Technology (Georgia Tech) announced today the launch of the Georgia Tech–Amazon Science Hub, reinforcing a shared commitment to advancing cutting-edge research through academic–industry collaboration. The Hub builds on extensive existing collaborations between Georgia Tech and Amazon over the past decade.

As the Amazon Research Awards (ARA) program continues to expand in scope, so does our call for proposals list. To that end, starting October 1, we will open the Fall 2026 call for proposals in nine research areas listed below.

Graphs encode relationships. In a world where AI agents need to reason about complex systems, not just retrieve information, graph structure provides the scaffolding for causal inference, dependency tracking, and compositional reasoning. Unlike tabular or unstructured data, graphs preserve the topology of real-world systems: what connects to what and how influence flows. For networks, this is not a metaphor. Networks are natively graphs.

When we fine-tune an LLM on code in a new programming language, what actually changes inside the model? Does it learn new algorithms, new syntax, or does it develop the ability to faithfully implement known algorithms in an unfamiliar language? This question is hard to answer with existing languages because we cannot disentangle what the model learned from fine-tuning versus what it already knew from pretraining.

Scaling test-time compute to search for optimal solutions is an important step towards building generally-capable language models that can reason. Recent work, however, shows that tasks of varying complexity require distinct search strategies to solve optimally, thus making it challenging to design a one-size-fits-all approach.

Modern Automatic Speech Recognition (ASR) systems are evaluated with respect to Word Error Rate (WER). While WER is a useful metric for training and evaluation of speech models, it does not fully reflect the difference in semantics between predicted and ground truth transcriptions. In conversational voice assistants, the ability to sufficiently understand the semantic meaning of the user request is often more important than recognizing all words correctly.

In this paper we introduce SemiGPC, a distribution-aware label refinement strategy based on Gaussian Processes where the predictions of the model are derived from the labels posterior distribution. Differently from other buffer-based semi-supervised methods such as Co-Match [17] and SimMatch [34], our SemiGPC includes a normalization term that addresses imbalances in the global data distribution while maintaining

Group relative policy optimization for reinforcement learning with verifiable rewards (RLVR) typically uses a fixed importance-sampling (IS) ratio clipping boundary across all rollouts. We identify a key limitation: rare correct rollouts on harder problems and abundant correct rollouts on easier problems are clipped at comparable rates, despite contributing very different learning signals.

Matching users to interest categories at scale is central to personalized shopping, but the task is challenging in large e-commerce platforms, where label spaces continually evolve and user-interest signals are sparse and long tailed. Autoregressive language models are appealing because their world knowledge and semantic priors over descriptors generalize across extreme label spaces and accommodate multiple valid label assignments.