ML, DL & AI Sales Forecasting Specialist

Hiring a Sales & Demand Forecasting Expert | ML, DL (Neural Networks), AI Project Overview We are looking for an experienced Machine Learning (ML), Deep Learning (DL - Neural Networks), and AI expert to help us develop a robust Sales & Demand Forecasting model. Our business deals with B2B wholesale distribution with over hundreds of SKUs across multiple warehouses. The biggest challenge? Our sales pattern is highly irregular, making traditional forecasting methods ineffective. What We Need We need a forecasting solution that can: Handle random sales patterns Accurately predict sales & demand fluctuations Integrate with our in-house software through API to pull real-time inventory & sales data Provide actionable insights to prevent inventory shortfalls & overstocking. (if possible, model will tell the reasoning behind the predicted numbers) Key Challenges & Requirements ✅ Sales data is NOT linear or seasonal – our sales pattern is highly irregular, meaning a simple time-series model like ARIMA or SARIMA won’t be effective. ✅ The solution must incorporate multiple real-world factors affecting demand. ✅ We have an in-house API that will provide real-time data – you need to work with our developers to connect the forecasting model to this API. ✅ High accuracy needed – the forecast should help improve inventory planning, supplier management, and backorder handling. Ideal Solution ML Models: XGBoost, LightGBM, Random Forest for multi-variable forecasting DL Models: LSTM, Transformer-based models for handling complex dependencies Hybrid Approach: Combining traditional statistical models + AI-driven approaches Nuera Link: (If applicable, let us know how you can integrate it into forecasting) Available Data Reports for Forecasting We have a comprehensive dataset that captures everyday inventory, sales, in-transit shipments, and customer buying behavior. These reports will be used to train the ML/DL model for better forecasting accuracy. ? 1. Inventory Levels Report (By Warehouse) - Last 3 Years Why? This will help identify trends, seasonal cycles, and fluctuations in stock movement across different warehouse locations. ? 2. Real-Time Current Inventory Levels Why? A real-time inventory snapshot helps adjust forecasting dynamically based on current stock availability to avoid overstocking or shortages. ? 3. In-Transit Inventory by Warehouse Why? Shipments on the way impact future inventory availability. This data will help predict when inventory will be available for order fulfillment. ? 4. Back Orders by Warehouse Why? Backorders indicate unfulfilled demand and missed sales opportunities. The model must adjust predictions to compensate for lost sales due to stockouts. ? 5. Daily Invoices by Warehouse Why? Analyzing daily sales patterns helps improve short-term demand forecasting. It also provides insights into customer buying frequency. ? 7. Inventory Shortfall Queries by Warehouse Why? Tracks how often products were searched for but not available, helping to forecast missed demand. ? 8. Item Ledger by Warehouse Why? Gives a detailed transaction history of each item across warehouses to track stock movements, adjustments, and shrinkage. ? 9. Inventory Turnover & Stockout Report Why? This helps determine which products sell fast vs. slow and adjust forecasts accordingly. ? Who We Are Looking For ✔ Proven experience in ML, DL (Neural Networks), and AI-based forecasting models ✔ Strong expertise in handling random, non-linear sales patterns ✔ Ability to connect the forecasting model to our API ✔ Experience with Python, TensorFlow, PyTorch, XGBoost, LightGBM, SARIMA, Prophet ✔ Understanding of warehouse inventory forecasting is a plus ⏳ Project Timeline & Deliverables ✔ Phase 1: Data analysis, preprocessing, and model selection ✔ Phase 2: Train ML/DL models using our historical sales & inventory data ✔ Phase 3: API integration with our in-house system for real-time predictions ✔ Phase 4: Model evaluation & accuracy improvement ⏳ Expected Completion: 3-4 weeks ? Budget ? Competitive – Open to discussion based on experience & deliverables. If you are confident you can develop a highly accurate forecasting solution, apply with: ✅ Your previous forecasting projects (preferably handling random sales patterns) ✅ Which ML/DL models you would use and why ✅ How you will integrate the model with our API