We advance science and technology to benefit humanity.
Joined February 2009
- Tweets10.2K
- Following1.8K
- Followers556K
- Likes5.8K
Microsoft Research retweeted
4B parameters. ZERO distillation. 61.5% on SWE-bench Verified 馃く
Meet FrogNano 馃惛: Qwen3.5-4B post-trained purely with RL on synthetic tasks from TaskPilot.
Just 5 iterations 脳 300 tasks.
Who said coding agents have to be huge? 馃惛
Curious about what a future in research could look like? The Microsoft Research Undergrad Internship Program offers hands-on research experience, mentorship, and a chance to explore advanced study. Rising juniors and seniors can apply for 12-week internships in Redmond, NYC, or New England.
Apply by October 5: msft.it/6014aXVWM
ALT Three people gather around a laptop at a shared desk, reviewing content together in a bright office environment with computer workstations in the background.
Where can curiosity take a research career?
Undergraduate research intern Matheus Kunzler Maldaner explores agentic systems and finds more than he expected in this first episode of our series, Research | Start Here: Inside the Internship.
See where research can start 鈻讹笍
We're thrilled to celebrate two Microsoft Research papers recognized at VLDB 2026.
Congratulations to the researchers. msft.it/6012aTYVm
Best Paper Award: Garnet: A Next-Generation Cache-Store for Accelerating Applications and Services
Garnet reimagines the cache-store layer that powers modern applications and cloud services, helping deliver orders-of-magnitude higher performance and efficiency at scale without requiring any application changes. The Garnet system is also available for use as open-source software. msft.it/6013aTYVW
ALT Best Paper Award: Garnet: A Next-Generation Cache-Store for Accelerating Applications and Services
Honorable Mention: Scalable GPU Acceleration of Scalar Functions in Analytical Databases: Compilation, Benchmarking, and Optimization
Analytical databases support hundreds of scalar functions that often become query bottlenecks. This work introduces a compiler-based approach that translates production database scalar functions into efficient GPU kernels while preserving their original semantics, enabling substantial performance gains for analytical workloads. msft.it/6014aTYVo
ALT Honorable Mention: Scalable GPU Acceleration of Scalar Functions in Analytical Databases: Compilation, Benchmarking, and Optimization
What if pathology foundation models could do more with less? GigaPath-Flash and GigaTIME-Flash cut computational demands while maintaining strong performance, opening the door to larger studies and broader exploration. msft.it/6012arcfr
We treat the way the world works now as normal. It isn't. Doug Burger, Amy Luers & Ishai Menache sit with a hard idea: modern civilization is a historical anomaly built on fossil fuels & exponential growth & AI may be the first tool capable of rewiring it. msft.it/6013azL6Z
Skala 1.1, the updated deep-learning exchange-correlation functional from Microsoft Research, provides greater accuracy, expanded accessibility across the computational chemistry ecosystem, and a living benchmark to track computational performance. msft.it/6011aOXe5
Every powerful new technology comes with trade-offs.
The internet had them.
Social media had them.
Doug Burger, Amy Luers, and Ishai Menache explore AI鈥檚 climate impact and steering the tech to support a sustainable future. msft.it/6010azEQA
Microsoft is proud to be a part of SIGCOMM 2026. Check out our sessions, covering network traffic engineering, new load balancing strategy, adaptive photonic switching schedules, AI network simulation, and more. msft.it/6018aKgfp
A path, a fence, a knot. MindTopo sets a new benchmark for testing how AI understands topological relationships and highlights new opportunities to strengthen spatial reasoning and planning. msft.it/6011aH02S
Radiology AI is evolving beyond report generation. CARE-X explores a unified approach that combines flexible reasoning, calibrated predictions, and measurement-based tools for chest X-ray interpretation. msft.it/6015aHrqH
The best AI doesn't just scale, it adapts to the languages, cultures, people and communities it serves. Explore three playbooks created to help teams design, deploy, and evaluate AI systems with inclusivity in mind.
ALT Two people sit side by side outside a building, examining information on a smartphone. One person gestures toward the screen as they review the content together.
Atlas: Frameworks and tools for designing and deploying human-centered AI systems across diverse cultural contexts. msft.it/6017aEk6D
Vibhasha: A practical guide for building multilingual and multicultural AI systems that are both linguistically accurate and culturally relevant msft.it/6018aEk6E
Can AI learn pathology through clinical dialogue? Introducing PRISM2, a multimodal foundation model trained on pathology images and language from real pathology reports. Using simple question-answering, PRISM2 matches specialized cancer-detection systems across several benchmark tasks without requiring a separate model for each task. msft.it/6017aDr1t
Small language models learn to negotiate with SocialRL, PazaBench V2 expands speech AI evaluation across African languages, and EvoLib helps agents turn experience into knowledge. Plus: new methods for more reliable A/B testing and advances in AI-driven precision oncology. msft.it/6012a8hjI
Orchard is an open-source framework for the research community to train and evaluate AI agents across task types. It reduces complexity while supporting strong performance from smaller models by enabling researchers to reuse the same infrastructure. msft.it/6019a8fqP