About Company
Data science company with a software platform that runs secure and compliant machine learning workloads at scale. We specialize in auditable environments and in synthetic data for defense use cases where the source data cannot leave a government installation. Our first program builds physics-informed failure data for aircraft drivetrain health monitoring, working with an engineering partner that models how bearings and gears fail.
THE ROLE
You will lead the data science on a program that generates physics-informed synthetic vibration data for drivetrain health monitoring. You will work with a partner's physics-of-failure models of bearings and gears, turn their fault signals into realistic sensor data, rebuild the condition indicators that fielded health monitoring systems compute, and prove with sound statistics whether the synthetic data improves fault detection. Compute is capped, so experiment design matters more than brute-force search.
WHAT YOU WILL DO IN THE FIRST SIX MONTHS
Build a signal processing library and an emulator of fielded condition indicators: time synchronous averaging, envelope analysis, band energy, shock pulse and kurtosis measures.
Fit a fast cyclostationary signal layer to physics model output for Monte Carlo generation.
Define fault severity levels and degradation branches with the physics team, including lubricant depletion and stalled growth.
Screen the physics models against the generated data and report which show promise.
Run a conditional diffusion model as a challenger and a feature-level domain adaptation model as the control.
Design and run the statistical significance demonstration: held-out splits by source, paired tests, ROC analysis and exact binomial bounds on detection and false alarm rates.
Write the prototype report and the technical sections of monthly reports.
REQUIRED
A PhD, or a master's degree with equivalent experience, in mechanical or aerospace engineering, electrical engineering with signal processing, applied mathematics or statistics.
Three or more years applying machine learning to vibration, rotating machinery or comparable sensor time series.
Working command of bearing and gear fault diagnostics: defect frequencies, envelope analysis, order tracking, cyclostationary signals.
Statistical validation: ROC analysis, confidence bounds, hypothesis testing, experiment design.
PyTorch and the Python scientific stack (NumPy, SciPy, pandas).
Clear technical writing for engineers and program managers.
PREFERRED
Helicopter health and usage monitoring (HUMS), condition-based maintenance, ADS-79 or seeded-fault testing.
Physics-informed machine learning, surrogate models (Gaussian process, polynomial chaos) or Bayesian calibration (PyMC, Stan).
Diffusion models or other generative models for time series.
Domain adaptation methods (DANN, MMD, CORAL).
Publications in PHM Society proceedings, Mechanical Systems and Signal Processing or comparable venues.
WORKING ARRANGEMENT
This is a remote role, open to candidates anywhere in the United States. We provide the hardware and a company-managed laptop.
CITIZENSHIP AND CLEARANCE
U.S. citizenship required. An active Secret clearance is a bonus; eligibility to obtain one required.
EEO:
Mindlance is an Equal Opportunity Employer and does not discriminate in employment on the basis of Minority/Gender/Disability/Religion/LGBTQI/Age/Veterans.