AI for Research Automation

The quest for answers to anything imaginable has fueled technological progress for centuries. As the complexity of underlying physical systems grew, so did the need for intelligent automation to scale and advance.

System

We are building a generalist control agent for diverse experimental tasks — scalable AI systems capable of learning meaningful representations from raw sensory data and interacting with real-world scenarios, to test the acquired knowledge — harnessing machine intelligence to accelerate discovery.

Expanding the horizons of scientific understanding, while building the foundation for self-driving experimental labs of the future.

Generative World Models

Imagine, plan, and adapt

Learn from raw sensory data

Quantized reinforcement learning

Sample Efficient

Pre-trainable feature learning

Optimal out-of-the-box performance

Real-time control across domains

Self-Driving Labs

Autonomous experimental control

Generalising across hardware

Discovery automation

Connect

Arindam Saha

Experimental Physicist · AI Researcher