Partner with the business to translate requirements into clear problem statements, KPIs, and experiment plans (A/B, holdout, backtests).
Design data & ML architectures on lakehouse/warehouse stacks (e.g., Oracle Exadata, Spark/Databricks; Snowflake/BigQuery/Redshift with open table formats like Iceberg/Delta/Hudi or equivalent ).
Build pipelines for ingestion, feature engineering, and training (batch & streaming) using Python + SQL with orchestration (Airflow/Prefect/Dagster).
Model using scikit-learn/XGBoost/LightGBM and PyTorch/TensorFlow; manage experiments and lineage
Serve & operate models on a major cloud ML platform (Azure ML, SageMaker or Vertex AI), with CI/CD, canary/blue-green, and rollback guardrails.
Monitor & improve: implement data/model quality and drift monitoring, alerting, and dashboards; close the loop with BI (Power BI/Tableau/Looker).
Document & review: author concise design docs and run technical reviews; mentor engineers; champion responsible AI practices.
Who you are & how you'll make your mark:
8+ years in applied ML & data engineering (3+ years leading delivery of production ML systems).
Python expert with production-grade SQL; strong with pandas/Polars, scikit-learn, and one of: XGBoost/LightGBM.
Fluency in core ML toolkits including TensorFlow, PyTorch, scikit-learn, and familiarity with Hugging Face or equivalent frameworks.
Proven record of constructing and maintaining scalable data pipelines—both batch and streaming—for model training and deployment.
Data platforms: hands-on with one of: Oracle ExaData, Spark/Databricks, or Snowflake, BigQuery/Redshift or equivalent; comfortable with open table formats (Iceberg/Delta/Hudi).
Orchestration: real projects using one of Airflow, Prefect, or Dagster.
Cloud ML platform: production deployments on one of SageMaker, Vertex AI, or Azure ML (pipelines, endpoints, registries).
MLOps: CI/CD for ML, experiment tracking, model registry, observability (latency, errors), and data/model drift monitoring.
Communication: ability to frame trade-offs and influence cross-functional partners; crisp writing of design/decision docs