Data Scientist

We are looking for a Senior Data Scientist with strong classical ML expertise to design, build, and operationalize predictive models within the
Microsoft Fabric
ecosystem. You will work on high-impact use cases spanning demand forecasting, risk scoring, and anomaly detection for large-scale retail environments — translating raw data signals into actionable business intelligence.

Key Responsibilities:

- Design and develop end-to-end classical ML pipelines — from feature engineering to model deployment and monitoring
- Build
demand forecasting models
leveraging external data signals (weather, events, seasonality) alongside historical sales data, at store/category/SKU level with 1–14 day horizons
- Develop
ML-based risk scoring
models across multiple fraud and exception scenarios, replacing manual rule-based processes with adaptive, dynamic thresholds
- Deliver daily prioritized outputs (investigation lists, inventory signals) that reduce detection and decision cycles from weeks to days
- Own model validation, threshold tuning, false positive reduction, and ongoing performance monitoring in production
- Collaborate with data engineers on feature pipelines using
Microsoft Fabric
Lakehouse, Dataflow Gen2, and OneLake
- Participate in iterative pilot-to-production delivery cycles with structured feedback incorporation
- Communicate model outputs and business impact clearly to both technical teams and business stakeholders

Required Skills & Experience:

• 8–12 years of hands-on Data Science experience with a strong foundation in classical ML
• Proficiency in supervised and unsupervised ML techniques — gradient boosting, regression, classification, anomaly detection, time-series forecasting (XGBoost, LightGBM, scikit-learn, Prophet, statsmodels)
• Strong hands-on experience with Microsoft Fabric — ML Experiments, Notebooks (Python/PySpark), Lakehouse, Pipelines, and Dataflow Gen2
• Solid Python programming skills with experience building production-grade ML code
• Experience with MLflow for experiment tracking, model registry, and lifecycle management (native within Fabric)
• Proven experience building time-series forecasting models at granular levels (store, SKU, or category)
• Experience with anomaly detection and risk/fraud scoring models in retail or financial domains
• Strong skills in feature engineering, cross-validation, model interpretability (SHAP, LIME), and drift detection

Nice to Have:

- Experience integrating
external data enrichment sources
(weather APIs, economic indicators, third-party signals)
- Familiarity with
retail loss prevention
, exception-based reporting, or shrinkage analytics
- Exposure to
Power BI
or Fabric-native reporting for operationalizing model outputs to business users
- Knowledge of
Azure ML
and its relationship with Microsoft Fabric ML capabilities
- Experience with irregular or non-reorderable inventory environments