Retail supply chains, energy utilities, FinTech treasury teams, and SaaS capacity planners frequently search to hire a freelance time-series forecasting (StatsForecast, Nixtla TimeGPT, Chronos, Prophet, LightGBM) & anomaly detection programmer. I am Aldian Fazrihady, a Senior Freelance AI & Data Engineer.
Core Time-Series Forecasting & Inventory Optimization Capabilities
- Hierarchical & Multi-SKU Demand Forecasting: Forecasting 100,000+ SKU-location series using LightGBM, Nixtla `StatsForecast` / `MLForecast`, Amazon Chronos / TimesFM foundation models, and hierarchical reconciliation so store-level forecasts sum accurately to regional totals.
- Probabilistic Prediction Intervals (P10 / P50 / P90): Computing safety-stock reorder points, lead-time uncertainty buffers, and promotional/holiday lift features.
- Automated Pipeline & Interactive Planner UI: Wrapping forecasting models in scheduled Dagster/Celery pipelines with React override dashboards for supply-chain planners.
Frequently Asked Questions
How do you measure and prove forecasting accuracy before going live?
I implement rolling-origin time-series cross-validation (backtesting across multiple historical cutoff dates) and compare WAPE, MASE, and RMSE against seasonal naive baselines.
How do I hire you for time-series forecasting & demand planning?
Hire me via Upwork.