Research

Advancing the frontiers
of technology.

Our research spans artificial intelligence, robotics software, and chip architecture — exploring ideas that push boundaries and create real-world impact.

Research Areas

Artificial Intelligence

Building intelligent systems that understand, reason, and assist — from language models to multimodal AI.

Large Language ModelsMultimodal ReasoningEfficient InferenceAI Safety

Robotics Software

Developing the software stack for autonomous robotic systems — perception, planning, and control.

Perception SystemsMotion PlanningSim-to-Real TransferMiddleware

Chip Architecture

Research into custom accelerator designs optimized for AI and robotics workloads.

Sparse ComputeNeural ISAEnergy EfficiencyHardware-Software Co-design
Working Papers

These are topics we are actively researching. No published papers yet — just honest work in progress.

Language Models·In progress

Efficient inference techniques for sparse mixture-of-experts language models

Robotics·In progress

Learning-based manipulation policies with sim-to-real transfer for robotic arms

Robotics·In progress

Real-time sensor fusion for autonomous navigation in unstructured environments

Hardware·Early research

Accelerator architecture exploration for sparse neural network workloads

Language Models·In progress

Multi-modal reasoning across text, vision, and proprioceptive sensor data