Machine Learning
Core ML: supervised, self-supervised, and reinforcement learning methods applied to real-world data and problems.
Research & Technology
Researching the building blocks
of intelligent systems
We work across the machine learning stack — from foundational research to the applied systems that put it to work.
AI & ML technologies
The core disciplines our research and engineering work spans.
Deep Learning
Neural network architectures and large-scale training systems, tuned for efficiency and stability at scale.
Large Language Models
Research into training, fine-tuning, and evaluating language and multimodal models.
Applied AI Systems
Turning research into deployable systems — inference infrastructure, tooling, and production ML pipelines.
Research projects
A look at the questions we're currently working on.
Efficient Long-Context Reasoning
Investigating architectures and training strategies that let models reason reliably over long, complex inputs.
Multimodal Representation Learning
Studying how models can learn shared representations across text, images, and other modalities.
Evaluation & Alignment Tooling
Building tools to measure model behavior, robustness, and alignment with intended goals.
Efficient Model Compression
Researching techniques to make capable models smaller, faster, and cheaper to run.
How we approach research
A disciplined process — from open question to shared result.
01
Explore
We start from open questions, not fixed roadmaps — following where the evidence leads.
02
Experiment
We build small, test fast, and let empirical results guide the next step.
03
Validate
Every result is stress-tested before we trust it — rigor over hype.
04
Share
Where we can, we publish and open-source what we learn.
Publications
We're a young research team. Our first public writeups are on the way.
No publications yet
When we have results worth sharing, we'll publish them here — along with code and data where we can. Reach out if you'd like to hear when that happens.
Get in touchJoin us
Want to work on problems like these?
We're a small, senior team looking for researchers and engineers who want to do their best work on hard AI problems.
See open roles