Combinatorial Design of Experiments Specialist

About the Role

Our benches and pilots have many knobs: sensor placement, sample rates, drift regimes, interference sources, room airflow states, calibration variants, pipeline configs, and acceptance thresholds. Trying every combination doesn’t scale, but choosing tests ad hoc creates blind spots that only show up in the field.

This role applies combinatorial design of experiments to create small, structured test plans that still cover high-order interactions. You make it possible to say: “We ran N cases, and we covered the important interactions—here’s the coverage receipt.”

What You’ll Own

• Minimal-run test plans for benches, commissioning workflows, and pilots (which scenarios to run and in what sequence).
• Covering-array generation under constraints: produce t-wise coverage while respecting feasibility constraints.
• Coverage measurement + receipts: quantify what interactions were covered and where gaps remain, with auditable artifacts.
• Experiment libraries: reusable scenario templates for common failure modes and operational conditions.
• Coordination with validation/field: ensure test plans align with acceptance tests, CI regression suites, and commissioning reality.

What You’ll Do

• Define factors and levels across domains (bench geometry, sensor placement archetypes, pipeline settings, environmental regimes).
• Choose interaction strength (t): decide when pairwise is enough and when specific high-risk clusters need 3-wise/4-wise coverage.
• Generate covering arrays (mixed-level, constraint-aware), including forbidden combinations and conditional factors.
• Sequence experiments to reduce setup churn and isolate confounds (group calibration changes, minimize geometry moves).
• Integrate plans into QA pipelines: convert matrices into executable checklists / automated runs, and store results as datasets with metadata.
• Iterate based on evidence: update factor levels and add targeted tests to close gaps without blowing up run counts.

Concrete Deliverables

• A factor catalog (by bench/pilot type): factors, levels, constraints, and rationale (“why this factor matters”).
• A covering array generator/tooling that outputs: test matrix, coverage stats, constraint-satisfaction receipt, suggested run order.
• Pilot test plans for cigar lounge modeler commissioning/validation, data center zone commissioning/alarm validation, and noise “air bubble” stability trials.
• A coverage dashboard/report: what interactions were covered, which failure modes are represented, what gaps remain.
• A minimal regression suite: a small set of “must-run” scenarios that complement property tests and guard core behavior.

Required Qualifications

• Strong experience in combinatorial testing / covering arrays / interaction testing (pairwise, t-wise, mixed-level designs).
• Ability to translate real systems into factors/levels without losing meaning or creating impossible matrices.
• Comfort with constraint handling and optimization in test generation (often CP-SAT/MILP-style thinking).
• Ability to build practical tooling in Python (common) with reproducible outputs and clear reporting.

Preferred Qualifications

• Experience designing experiments in physical systems (instrumentation, sensors, facilities/HVAC, RF/audio benches).
• Knowledge of common combinatorial design constructions (orthogonal arrays, Latin squares) and when they apply.
• Experience integrating experiment plans into CI/QA pipelines and dataset versioning.
• DSP/measurement intuition helpful for shaping realistic failure-mode factors (not required).

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

• You deliver a first minimal-run test plan the team actually uses, reducing bench/pilot time while increasing coverage confidence.
• You produce coverage receipts that make gaps obvious and discussions concrete.
• At least one non-obvious interaction failure is discovered early because it was included by the design (not by luck).
• Test plans become repeatable assets (templates + tooling), not one-off spreadsheets.

Working Style

• You think in interactions, not single knobs.
• You’re allergic to both brute force and vibes—you want structured minimality with receipts.
• You communicate clearly: “Here’s what we covered, here’s what we didn’t, here’s why.”

Title & Level

Combinatorial Design of Experiments Specialist (senior IC enabling role; can scale to Staff with ownership of experimentation strategy), partnering with validation, systems, and field teams.