Retail Predictive Analytics Data Scientist

I need a data scientist who can build a robust predictive analytics solution for a retail-focused use case. My datasets include sales transactions, product metadata, marketing campaign logs, customer reviews, and social media chatter, so you will be working with a mix of structured tables and unstructured text.

Your mission is to extract the signals that drive sell-through, forecast demand at SKU and store level, and surface actionable insights for merchandising and marketing. I expect you to handle everything from data ingestion and cleaning through to model deployment, with clear documentation of assumptions and feature engineering steps.

You may use Python (pandas, scikit-learn, XGBoost, Prophet, TensorFlow, or similar), SQL for warehousing, and NLP libraries such as spaCy or transformers for the text components. If you prefer R or another stack, I’m open as long as the final model meets the accuracy and interpretability goals.

Deliverables
• Cleaned and well-structured datasets ready for modelling
• Reproducible notebooks / scripts showing EDA, feature engineering, and model training
• A predictive model (or ensemble) with validated performance metrics
• Summary report and slide deck translating findings into retail business actions
• Deployment-ready code or API endpoint for ongoing predictions

Acceptance criteria
• Minimum MAPE < 10 % on the hold-out test set
• Code passes peer review for readability and modularity
• All steps reproducible on my environment using a requirements.txt or environment.yml

Timeline is flexible within reason, but please outline milestones for data prep, modelling, validation, and final delivery when you respond. I’m happy to answer any clarifying questions before we kick off.