What

Work in UC San Diego’s Integrated Systems Neuroengineering Lab on a Python algorithm that took neural spikes generated from simulated retina-like camera data and mapped them back to the visual phenomena that produced them.

Why it mattered

A retina-like sensor emits events, not pictures. Whether those events carry enough information to recover what was actually in front of the sensor is the question that decides whether the whole approach is viable — and it is a decoding problem, not an optics problem.

Personally it was the pivot point: first exposure to Python and to neural data, simultaneously. Every role since has been some larger version of those same two things.

My role

Improved the lab’s existing Python mapping algorithm.

Outcome

An improved implementation of the mapping algorithm, and the toolkit — Python, spike data, signal-to-meaning — used in every role that followed.