Research
General-purpose cryptographic proof systems enable the following. A program runs, and along with its output it produces a short proof, a cryptographic certificate that this program, on these inputs, produced this output. The proof is…
A Text-Based Environment for Interactive Debugging In most existing approaches (shown in black), an agent rewrites its code conditioned on error messages obtained from executing its code. debug-gym equips the agent with additional tools such…
the Text Adventure Learning Environment Suite Reasoning is an essential skill to enable Large Language Models (LLMs) to interact with the world. As tasks become more complex, they demand increasingly sophisticated and diverse reasoning capabilities…
We explore AI as a living medium: generative, malleable, and alive. By reimagining how people interact with intelligence, we seek to amplify human cognition, creativity, and problem solving.
Large language model (LLM) inference has become an operational workload that fleets must provision, monitor, and bill against latency service-level objectives (SLOs). A single throughput number cannot answer the questions operators actually face, because generative…
Tree-structured prefix sharing is now standard in large language model (LLM) serving: a radix-tree KV cache lets multiple requests reuse common context prefixes and reduces memory capacity pressure. Yet the attention kernels used with these…
Unlocking new gaming experience with agentic AI Microsoft VEGA TeamMicrosoft Research & XBOX What if, instead of controlling every move your game character made, you could tell them what you wanted and watch them live,…
Imitation learning enables agents to learn complex behaviour from demonstrations, but in practice it often requires large datasets that are costly or impractical to collect. Our project studies how to make imitation learning significantly more…
CARE is a research program at Microsoft Research India focused on building foundation and multimodal models for healthcare, with an emphasis on radiology. The program develops vision-language and representation-learning systems that prioritize clinical fidelity, interpretability,…
Project Sico (Symbiotic Intelligence for Co-evolution) investigates how agentic systems achieve persistent, system-level improvement. The key research theme behind Sico is agentic evolution, the shift from self-improving agents to co-evolving human–AI systems. Its central argument…