Systems Engineer (Bench → Contract)

About the Role

We build instruments and algorithms that only work when their assumptions are true on a given window of time. The Systems Engineer (Bench → Contract) turns a physical bench (or field deployment) into a set of explicit, testable promises: what the bench can guarantee, what the algorithm is allowed to assume, how we detect violations, and what “pass/fail” means before we let anyone trust a plot.

Think of this role as: take messy reality → produce contracts that software can enforce.

What You’ll Own

• Define “E/Q promises” (Experiment / Quality promises): write down what a bench must guarantee for algorithms to be valid (sampling, timing, clock behavior, placement, excitation conditions, constraints).
• Bench instrumentation for window truth: specify and implement the telemetry needed to score a window (W): coherence, drift, saturation/clipping, noise floor, stationarity, multipath/interference flags.
• Failure mode taxonomy: enumerate how benches fail in practice (and how it looks in data), then encode detection + labels that are actionable.
• Acceptance tests & go/no-go gates: create objective criteria and automated tests that decide whether a run is valid, calibration is stale, a window is usable, and results can be reported.
• Bench → pipeline integration: ensure metadata (geometry, sensor IDs, firmware/config versions, calibration constants) flows into the processing pipeline and is recorded as a receipt.
• Debug playbooks: build repeatable procedures for diagnosing failures (what to check first, what plots/metrics matter, how to reproduce).
• Cross-team translation: keep contracts realistic and enforced across hardware, algorithms, and product.

Concrete Deliverables

• A Bench Contract Spec per bench type: inputs (signals + metadata), required promises, allowed algorithm assumptions, mandatory diagnostics, pass/fail acceptance criteria.
• A window scoring + labeling module that emits: usable / borderline / invalid + reasons.
• A failure-mode library with examples: “what it looks like,” “how to detect,” “what to do next.”
• An acceptance test suite that runs automatically on new data and blocks regressions.
• A run receipt format (manifest) that makes every result reproducible.

Required Qualifications

• Strong systems thinking: define interfaces and guarantees between messy physical systems and software.
• Experience with measurement systems: sensors, DAQ, timing/synchronization concepts, calibration workflows, environmental effects.
• Comfort with data-driven debugging: interpreting spectra/coherence plots, diagnosing drift, spotting instrumentation artifacts.
• Ability to write clear specs and enforce them through automated checks.

Preferred Qualifications

• Experience with audio/RF timing, distributed sensing, synchronization, or noisy instrumentation.
• Familiarity with DSP basics (FFT, coherence, windowing/leakage) sufficient to reason about algorithm assumptions.
• Experience designing QA / verification regimes: acceptance tests, regression datasets, reliability metrics.
• Hands-on bench experience: you’ve built or operated rigs and know what “works in theory” misses.

How You’ll Be Measured (First 60–90 Days)

• You produce a first Bench Contract that the team actually uses to gate results.
• You implement (or drive implementation of) window validity instrumentation that catches common bench failures early.
• You deliver an acceptance test harness that converts "arguing over plots" into objective pass/fail outcomes.
• You reduce rework by making failures diagnosable and repeatable ("we can reproduce it and we know what broke").

Working Style

• You prefer explicit promises over vague optimism.
• You assume every bench will betray you eventually—and you design the system to detect and survive it.
• You're happiest when the pipeline can say: "This result is invalid, and here's exactly why."

Title & Level

Systems Engineer (Bench → Contract) (mid-to-senior; can scale to Staff depending on ownership), partnering closely with applied math/DSP and experimental operators.