What we're building

Autonomous agents that write, execute, and debug scientific code across domains — materials science, computational biology, and beyond.

An AI system that proposes, schedules, and analyzes physical experiments in a closed-loop pipeline with minimal human input.

A Python API for commanding robotic actuators using natural language instructions translated to low-level motor signals via LLMs.

A low-cost, 3D-printable 6-DOF robotic arm designed for AI control experiments, sensor integration, and rapid hardware iteration.

Soft and rigid gripper designs produced with FDM 3D printing, featuring embedded force sensors for feedback-driven grasp control.

A unified platform for fusing data from multiple sensor modalities — depth, force, IMU, and vision — into a real-time robot state estimate.

A conversational interface for controlling robots with spoken or typed natural language, with explainable action feedback to the operator.

An augmented reality layer that visualizes robot intent, sensor data, and AI model predictions over the physical workspace in real time.

Open-source tools for inverse design, property prediction, and autonomous synthesis planning in computational materials science.