Tech Stack
AirflowAmazon RedshiftBigQueryCloudPythonScikit-LearnSQL
About the role
- Design, train, and deploy production ML models, with emphasis on tree-based methods (XGBoost, LightGBM, CatBoost).
- Contribute to feature engineering and model evaluation approaches that balance accuracy, interpretability, and speed.
- Translate customer needs and product requirements into practical modeling tasks (classification, regression, ranking, uplift).
- Develop custom performance metrics to evaluate models’ ability to generate reasonable counterfactual demand estimates while accounting for varying customer product expectations.
- Support explainability efforts to justify rate decisions with clear artifacts and documentation.
- Build and maintain training and evaluation pipelines, ensuring offline metrics align with online business outcomes.
- Partner with engineering to integrate models into production services and batch jobs with defined SLAs.
- Contribute production ready Python and SQL code with strong testing and review practices.
- Support data quality, observability, and reproducibility across the ML lifecycle.
- Collaborate on experiment design, including A/B testing and staged rollouts.
- Implement policy guardrails to ensure automated rate changes are safe and controlled.
- Analyze experimental results, communicate lift and trade-offs, and document findings.
- Work closely with product, engineering, and leadership to scope problems and propose modeling approaches.
- Participate in code reviews, model reviews, and planning sessions.
- Share learnings and contribute to improving team practices around modeling, metrics, and documentation.
Requirements
- 4+ years of experience building and deploying ML models to production.
- Proficiency in Python (scikit-learn, XGBoost/LightGBM/CatBoost).
- Strong SQL skills and comfort with large, imperfect, real-world datasets.
- Experience across the full model lifecycle: data preparation, feature engineering, training, evaluation, deployment, and monitoring.
- Familiarity with experimentation design and metrics; able to reason about trade-offs and safeguards.
- Clear communicator, comfortable working in a collaborative, fast-paced environment.
- (Nice to have) Experience with causal inference (uplift modeling, causal forests, DML) or time-series modeling.
- (Nice to have) Exposure to reinforcement learning concepts (multi-armed bandits, dynamic programming, temporal-difference learning).
- (Nice to have) Experience with cloud data warehouses (Snowflake, BigQuery, Redshift) and dbt.
- (Nice to have) Familiarity with orchestration frameworks (Prefect, Airflow).
- (Nice to have) Familiarity with MLOps frameworks (Sagemaker, Vertex AI).
- (Nice to have) Domain knowledge in pricing, demand modeling, or hospitality.
- (Nice to have) Former start-up experience.