The Systems Planning and Architecture (SPARC) organization within Azure Hardware Systems & Infrastructure (AHSI) manages Azure’s hardware roadmap from architecture concept through production, helping power Microsoft’s “Intelligent Cloud” mission. During this Research Internship, you’ll focus on applied research and prototyping: how to use large language…
In the future of work discourse, AI is touted as the ultimate productivity amplifier. Yet, beneath the efficiency gains lie subtle erosions of human expertise and agency. This paper shifts focus from the future of work to the future of workers by navigating the AI-as-Amplifier…
The AI Frameworks team at Microsoft develops Artificial Intelligence (AI) software that enables running AI models everywhere, from world's fastest AI supercomputers, to servers, desktops, mobile phones, Internet of Things (IoT) devices, and internet browsers. We collaborate with our hardware teams and partners to build…
Do you come alive when a path isn’t obvious? Would you see open, complex, scientific, AI/ML, and engineering problems as something to be shaped into clear plans, crisp milestones, and aligned teams? We’re Microsoft Research Health Futures. A fast moving, global group of scientists, engineers,…
The Security Research Group at Microsoft Research Redmond is pushing the boundaries of data‑driven security. In this role as a Senior Researcher - Security, you will work alongside leading researchers and engineers to design and build next‑generation intrusion detection systems. Our work combines security expertise…
AI can help generate medical image reports, but today’s models struggle with varying reporting schemes. Learn how UniRG uses reinforcement learning to boost performance of medical vision-language models.
When humans face problems beyond their immediate capabilities, they rely on tools, providing a promising paradigm for improving visual reasoning in multimodal large language models (MLLMs). Effective reasoning, therefore, hinges on knowing which tools to use, when to invoke them, and how to compose them…
In LLM inference, the same prompt may yield different outputs across different runs. At the system level, this non-determinism arises from floating-point non-associativity combined with dynamic batching and GPU kernels whose reduction orders vary with batch size. A straightforward way to eliminate non-determinism is to…
Equal access to digital technologies is critical for education, employment, and social participation. However, mainstream interfaces are visually oriented, creating steep learning curves and frequent obstacles for screen reader users, and limiting their independence and opportunities. Existing support is inadequate -- tutorials mainly target sighted…
Multimodal Small-to-Medium sized Language Models (MSLMs) have demonstrated strong capabilities in integrating visual and textual information but still face significant limitations in visual comprehension and mathematical reasoning, particularly in geometric problems with diverse levels of visual infusion. Current models struggle to accurately decompose intricate visual…
In the arid expanse of northwestern Kenya, the Kakuma refugee camp has grown into a sprawling community of more than 300,000 displaced individuals from over 20 countries. Originally established in 1992 to shelter young people fleeing the war in Sudan, the refugee camp is the…
Frontier models have demonstrated remarkable capabilities in understanding and reasoning with natural-language text, but they still exhibit major competency gaps in multimodal understanding and reasoning especially in high-value verticals such as biomedicine. Medical imaging report generation is a prominent example. Supervised fine-tuning can substantially improve…
