ML Forecasting Engineer

Role: ML Forecasting Engineer

Experience: 1-3 Years

Job Location: Noida (Onsite)

About The Role

We are looking for an ML Forecasting Engineer to design, evaluate, and monitor predictive models that support high-impact, data-driven decision workflows. The role focuses on forecasting, scoring, ranking, model explainability, validation design, drift detection, and evaluation systems. The ideal candidate can translate ambiguous business questions into reliable machine learning designs while keeping privacy, bias, fairness, and model risk visible.

Requirements

Education Qualification

Bachelor’s / Master’s degree in a Data Science/ Artificial Intelligence/ Computational Mathematics/ Statistics or related field.

Required Skills

• Strong practical experience with forecasting, predictive modeling, ranking, scoring, or applied machine learning systems.
• Good understanding of backtesting, holdout design, validation methodology, calibration, and drift detection.
• Ability to reason about feature quality, leakage, proxy features, bias, explainability, and model monitoring.
• Experience working with structured, time-series, geospatial, marketplace, operational, or behavioral datasets.
• Working knowledge of production ML workflows, evaluation harnesses, model observability, and reproducible experimentation.
• Clear documentation skills for model design, assumptions, limitations, evaluation results, and monitoring requirements.
  
  

Responsibilities

• Design predictive models for forecasting, scoring, ranking, and decision-support use cases.
• Define feature families across structured, temporal, operational, behavioral, geographic, and external data sources without exposing sensitive implementation details.
• Build backtesting, holdout, and validation strategies that measure model quality before deployment.
• Develop evaluation plans covering accuracy, calibration, ranking quality, stability, data freshness, and business usefulness.
• Design explainability approaches so product, business, and governance stakeholders can understand model outputs and drivers.
• Monitor deployed or production-bound models for drift, degraded performance, data quality issues, and changing cohort or regional patterns.
• Identify and reduce risks from data leakage, proxy variables, privacy-sensitive features, bias, and unfairness across cohorts or regions.
• Partner with data, product, platform, and governance teams to make models reliable, auditable, and production-ready.
  
  

Related