We build products with AI training and inference at their core, engineered for performance and correctness.
We design and build scalable, fault-tolerant parallel and distributed systems for large-scale environments.
We investigate new approaches to hard problems in AI, security, and systems performance.