Start with light
An optical instrument is a demanding place to test intelligence. The physics is structured, the measurements are noisy, and a useful decision has a deadline. The atmosphere changes while the sensor exposes. A calibration ages. A mirror applies a command to a wavefront that has already moved.
Adaptive optics makes these problems measurable. We can ask how much light reaches the intended focus, how much wavefront error remains, and whether a more capable controller earns its computational cost.
Prediction needs a physical model
We build differentiable models because they let us trace how an optical parameter changes a measurement. They connect calibration, estimation, and control. But a differentiable model still needs to agree with independent physics, and a detailed model still needs to finish in time.
That is why our work begins with optical comparisons, derivative checks, and latency measurements. Those checks establish the basis for an experiment; controller comparisons then test whether the model improves the decision.
Autonomy needs a record
To understand a failure, we need more than the requested command. We need the exposure that informed it, when the observation became available, which calibration applied, and what the actuator actually did.
We are building telemetry qualification alongside control research because it makes those questions explicit. A recording with missing timestamps should produce an account of missing evidence, not an invented sequence of events.
The instrument is the destination
Our mission is to make optical systems operate, adapt, and recover autonomously. We begin with wavefront control and extend to optical receivers, where acquisition, tracking, coupling, and recovery must work together.
The sequence is deliberate: check the model, compare the control, qualify the telemetry, and validate the behavior on an instrument. Every stage should leave something another researcher can inspect.
If you work on the optics, hardware, or control problems along that path, we would like to hear from you.
Salma Shaik
Founder & CEO, Noether Labs