Senior ML Engineer (Energy & Utilities)

Responsibilities

• Own the reusable utility ML libraries — forecasting, disaggregation, demand response, detection, and asset health — each shipped with its own evaluation harness and documentation.
• Build capabilities once as tested libraries with validation harnesses, so client pods deploy proven components (like meter disaggregation or load forecasting with abstention and monitoring built in) instead of reinventing them per engagement.
• Own the simulation engine — Rust crates and Python bindings — including its validation methodology against the reference oracle and its performance, and use it to make city-scale building stock tractable through representative-archetype simulation.
• Own the grid-data toolkits: protocol codecs (IEEE C37.118-class), synthetic scenario generators, and verification utilities, including generating synthetic-but-believable meter data calibrated until domain experts can't tell.
• Build the planning and dispatch support behind demand-response programs, where acting on a wrong number carries real cost.
• Own the publish path — versioned crates and Python packages — shipped with the evidence (evals, benchmarks) attached.
• Own the feedback loop with client pods: track what the capability got wrong in the field, and turn that into what you build next.

Core Qualifications

• 5+ years of shipping applied ML on real-world signals — forecasting, disaggregation, detection/classification on interval or sensor data, survival/reliability modeling, or an adjacent-industry equivalent
• Strong numerical and scientific computing skills: feature engineering from raw interval data, solver-level numerics when needed, and a healthy distrust of your own metrics.
• Python plus a systems language — the models and tooling are Python, the engine is Rust; depth in one, working ability in the other, and the appetite to close the gap (Rust is teachable here; modeling judgment isn't).
• Library craft: you build things other engineers consume — versioned, tested, documented, with an API you'd want to call yourself.
• Willingness to do your own data engineering — finding, cleaning, joining, and profiling inputs yourself rather than trusting a prepared dataset.
• Practical fluency in our core stack — Python 3.12+ (numpy, pandas/polars, scikit-learn, statsmodels), time-series feature engineering, forecasting/clustering libraries (sktime/statsforecast-class), and SQL/Postgres or TimescaleDB-class hypertables.
• Comfort picking up Rust (or a comparable systems language) via PyO3/maturin, and building evaluation harnesses and CI for scientific software.
• Willingness to ramp quickly on energy-domain vocabulary if you don't already have it
• Bachelor's Degree; Master's is a plus

Why AZX!

Additional Information

Next Steps