QuDaQ

QuDaQ was the data-acquisition and instrument control layer behind a quantum benchmarking effort at NIST Boulder. The job was unglamorous and essential: turn noisy, hardware-level measurements into numbers you can actually trust and reason about.

The problem

Characterizing a quantum processor means running a lot of carefully constructed circuits, collecting shot statistics, and fitting them to models of how the hardware misbehaves. The hard part isn’t any single step — it’s that every step leaks error. Calibration drifts between runs. Fit routines quietly diverge on edge cases. A clean-looking decay curve can hide a systematic bias that makes the final fidelity number meaningless.

The approach

I worked on the acquisition and analysis pipeline that sat between the control stack and the results: generating randomized-benchmarking and randomized-compiling sequences, orchestrating their execution, and fitting the resulting decays to extract error rates. The emphasis throughout was on defensible numbers — surfacing when a fit was unreliable rather than reporting a confident-looking value, and keeping the path from raw shots to published metric legible enough that someone else could audit it.

The instinct I took from this: a measurement you can’t trace back to its assumptions isn’t a measurement, it’s a vibe.

What came out of it

A reusable characterization workflow that made the team’s benchmarking runs reproducible and easier to interrogate. {TODO: Add a concrete result here — e.g. the specific protocols supported, a throughput improvement, or a publication / internal report this fed into.}