Test & Validation Engineer

About the Role

We build systems that can look correct right up until the day they don’t. A notch that used to dip stops dipping. A “shared component” suddenly won’t cancel. A firmware update changes timing by 20 ppm and nobody notices until a customer does.

The Test & Validation Engineer makes this impossible to ignore by turning bench reality into repeatable rigs + golden tests + CI gates. Your job is to ensure that when something breaks, the system tells us exactly what broke—automatically—before it ships.

What You’ll Do

• Own bench validation rigs: design and operate test setups that produce known, repeatable stimuli (injection signals, loopbacks, controlled delays/drifts, controlled RH/T steps where relevant).
• Golden tests & reference artifacts: create canonical datasets (“goldens”) and expected outputs (metrics + plots + summaries) that become the regression baseline.
• Failure-mode-driven diagnostics: encode the team’s intuition as executable rules: “If notch does not dip…”, “If coherence collapses…”, “If phase slope becomes unstable…”, “If drift tracker diverges…” with likely causes + next checks.
• CI integration: wire validation into CI so algorithm/firmware/config changes can’t silently degrade performance.
• Metric definitions: define pass/fail thresholds (notch depth, residual energy, delay error, false-accept rates, abstain rates, runtime limits).
• Reproducibility receipts: ensure every test run logs code versions, configs, dataset hashes, and environment info so failures are reproducible.
• Triage + escalation: build “first responder” workflows—when CI fails, it’s obvious what to look at and who should fix it.

Concrete Deliverables

• A bench rig suite (starter rig + roadmap): injection + loopback + controlled delay/drift + noise/interference modes.
• A golden dataset pack with ground truth, metadata, and clear expected signatures.
• A CI validation pipeline that outputs: pass/fail + metrics report + diagnostic hints + artifact bundle (plots/logs).
• A “notch health” dashboard (even simple) showing regressions over time and which change introduced them.
• A failure-mode playbook: symptoms → likely causes → confirming tests.

Required Qualifications

• Strong testing mindset: you’ve built regression suites for systems where correctness is statistical/signal-based, not just unit tests.
• Experience with instrumentation/bench work: you can design rigs, run controlled experiments, and debug the physical layer when signals look wrong.
• Comfort with DSP-adjacent metrics (FFT/Welch, coherence, cross-correlation, SNR/noise floor, drift) enough to write meaningful tests.
• Ability to implement automation in Python (and/or similar) plus CI tooling (GitHub Actions, GitLab CI, etc.).

Preferred Qualifications

• Experience testing algorithmic systems where outputs are distributions (tolerances, confidence, abstention behavior).
• Familiarity with hardware-in-the-loop testing, lab automation, embedded validation, or synthetic data generators.
• Strong documentation chops: turning tribal knowledge into checklists and automatic diagnostics.

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

• You stand up a first end-to-end validation pipeline (dataset → run → metrics → pass/fail → artifacts) in CI.
• At least one major regression gets caught in CI that otherwise would have shipped.
• “Notch does not dip” becomes a diagnosed failure with actionable labels, not a vague complaint.
• Bench rigs produce repeatable results the whole team trusts as ground truth.

Working Style

• You prefer “make it fail loudly” over “hope it stays fine.”
• You believe every key plot should have a metric behind it and a gate in CI.
• You enjoy turning messy, real-world failure stories into deterministic tests.

Title & Level

Test & Validation Engineer (mid-to-senior; can scale to Staff if owning the validation architecture), partnering closely with systems, firmware, and DSP teams.