Machine Learning Engineer

ServiceUp

5d•Remote

About The Position

You'll own the full lifecycle of machine learning at ServiceUp, from exploration and model development through production deployment and iteration. You'll collaborate closely with product, design, and engineering to identify where ML can create value and translate model outputs into features customers use. Our data stack today includes PostgreSQL, BigQuery, and Hex, but this is our first dedicated ML hire. We'll look to you to help shape the ML tooling and infrastructure.

Requirements

7+ years of applied ML experience with a track record of shipping models to production.
Strong statistics and ML fundamentals. Bias-variance tradeoffs, class imbalance, calibration, and knowing when a simpler model wins.
Deep experience with tabular data, feature engineering on structured data, and messy real-world datasets.
Production ML experience. Feature drift, data quality, monitoring, and the gap between offline metrics and real-world performance.
Proficiency with ML tooling and frameworks. Our stack is Python-based today, but we care more about shipping production-quality code than any specific language.
SQL fluency for data exploration, feature building, and model validation.
Excellent communication and collaboration skills in a remote environment.

Nice To Haves

Domain experience in marketplaces, fleet operations, or repair/maintenance workflows.
Anomaly and fraud detection experience.
Practical LLM experience for explanation generation or data enrichment (not as the core model).

Responsibilities

Own and expand the model portfolio.
Drive the ML engineering process from idea to production.
Define success metrics and evaluation frameworks for each model.
Productionize ML models. Take experiments to production with real-time inference, monitoring, retraining pipelines, and phased rollouts with feature flags and guardrails.
Build and evaluate classifiers with rigorous offline and online methods.
Design feature engineering pipelines across repair orders, shop performance, fleet behavior, vehicle attributes, and cost benchmarks.
Own model evaluation including cross validation, XAI, confidence calibration, bias detection, A/B testing, and shadow mode comparison against human decisions.