Learn about Amazon's scientific research, science community, and career opportunities in artificial intelligence (AI), machine learning (ML), computer vision, robotics, quantum, economics and more.
Ten years after we founded the Automated Reasoning Group, mathematical logic has moved from academic research into production services that secure millions of customer workloads demonstrating that systems can be provably correct, not just probably correct.
A competition with a finalist ceremony during NeurIPS 2026, challenging researchers to train language models from scratch on Trainium, exploring what optimal architectures look like when the hardware changes.
Amazon announces 34 recipients of the Build on Trainium program, a $110 million credit initiative supporting AI research at 30 universities including Stanford, UC Berkeley, UIUC, UCLA, CMU, and MIT, with a focus on Responsible AI.
Instead of compromising among parameter updates dictated by different training objectives, ControlG allocates computational capacity to objectives sequentially and dynamically.
PatientAgentBench generates a synthetic patient health record, a realistic clinical vignette, and a patient agent that converses with the AI system under evaluation, to capture what a patient-facing agent actually has to do.
As AI agents take on higher-stakes decisions, Lean programming language makes it possible to mathematically prove they will behave safely.
HydroShear, a new physics-based simulator, teaches robots how to use their sense of touch to perform complex manipulation tasks, in a way that transfers seamlessly to the real world.
A new Rust proxy called Turnstile sits between the model backend and the agent harness to capture information lost in mere text transcripts.
Amazon is developing precise, sector-specific approaches to measuring decarbonization progress starting with emissions per unit shipped.
Millimeter-scale particles of nuclear-reactor fuel are encased in four layers of different materials that act as a miniature containment system .
Splitting the separation kernel off from the rest of the Nitro security system and using only a subset of the Rust programming language to code it enabled its formal verification.
A new chiplet architecture, custom die-to-die connectivity, and support for DDR5-8800 memory and the latest PCIe gen6 interconnects improve performance by 25% for general-purpose and agentic AI workloads.
Four approaches can dramatically improve the performance and trustworthiness of AI agents in operational environments.
The harnesses that mediate between models and tools in agentic systems are becoming their own performance bottleneck, but a few simple design principles can fix what ails them.
Automatically fact-checking long, AI-generated research reports poses new challenges including benchmarking.
Quasi-random network topologies and new passive optical components called ShuffleBoxes make more-efficient flat networks as practical as traditional fat-tree networks.
Awardees represent more than 49 universities in 11 countries. Recipients have access to Amazon public datasets, along with AWS AI/ML services and tools.
How to train language models to generate diverse, accurate reasoning paths using tokens that control distinct reasoning strategies.
A new scaling law that relates particular architectural choices to loss helps identify models that improve throughput by up to 47% with no loss of accuracy.
By focusing on specific failure points and suggesting targeted solutions, a new automated prompt-engineering framework improves prompt performance without compromising existing functionality.
Amazon engineers and scientists have created new tools to optimize delivery networks under uncertainty and keep them adapting without missing a beat.
Agentic mechanism enables Amazon and vendors to optimize supply chain management without disclosing private information.
Amazon scientists and policy experts discuss how the company s responsible-AI pipeline embeds safety and values throughout the AI development lifecycle.
How we reproduced three attacks that extract private training data from AI models and the cryptographic defenses that stop them.
A new framework provides a statistical method for estimating the likelihood of catastrophic failures in large language models in adversarial conversations.