Centram Centre for Applied Artificial Intelligence

AI and Machine Learning Course for Procurement, Supply Chain & Logistics Management

Course title;  AI & Machine Learning for Procurement, Supply Chain & Logistics Management
Target audience

Supply chain, procurement, logistics or operations managers and analysts
Data scientists and engineers working with supply‑chain teams
Consultants and product managers building supply‑chain solutions
Assumes basic analytics familiarity; adaptable for non‑technical executives (executive variant)
Prerequisites
Basic statistics, Excel; recommended Python or R (or plan a pre‑course refresher)
Familiarity with supply‑chain concepts (inventory, lead time, SKUs, BOM, routes)

Course learning objectives

By the end of the course participants will be able to:

1. Identify high‑impact AI/ML applications in procurement, supply chain and logistics.
2. Prepare, feature‑engineer and validate time series and transactional supply‑chain data.
3. Build and evaluate forecasting and classification models tailored to demand, lead time, and supplier risk.

4. Formulate and solve optimization problems (inventory policy, routing, allocation).
5. Use simulation and reinforcement learning for operational decision support.
6. Extract value from unstructured procurement data (contracts, invoices, images).
7. Deploy models in production workflows, monitor drift, and integrate outputs into decision dashboards.

8. Address ethics, supplier fairness, procurement governance, and data privacy.
Course format & time commitment

4 Weeks boot camp OR 2 Weeks physical and 3 weeks online training

Delivery: in‑person or online (Colab/GitHub + Zoom + dashboards).

Overview: AI opportunities & problem framing

  1. Topics: value chain from procurement to delivery, common use cases and ROI (demand planning, inventory, supplier risk, freight optimization). How to frame ML problems (prediction vs optimization vs rules).
  2. Lab: Map use cases to your organization or a case study (e.g., retail, manufacturing, last‑mile).
  3. Assignment: 1‑page problem statement with KPIs and data availability.

Data architecture & quality for supply chains

  1. Topics: ERP, TMS, WMS, e‑procurement logs, POS, IoT sensors; master data issues (SKU hierarchies, units); missing data, timezone/seasonality, data lineage.
  2. Lab: Clean and harmonize SKU and transaction data, create time series for SKUs by location.
  3. Assignment: Data readiness checklist and plan to improve master data.

Demand forecasting & baseline methods

  1. Topics: forecasting fundamentals, seasonality, intermittent demand (Croston), hierarchical forecasting, cross‑SKU/causal features, evaluation metrics (MAE, MAPE, WAPE, sMAPE).
  2. Lab: Build baseline forecasts (moving average, exponential smoothing, Prophet) and evaluate on retail/sales dataset (e.g., M5 or Rossmann).
  3. Assignment: Forecast for a product family; produce error analysis and baseline recommendations.

Machine learning forecasting & probabilistic forecasts

  1. Topics: tree models (LightGBM/XGBoost) for forecasting, feature engineering (lags, rolling stats, promotions, price, calendar), quantile regression and prediction intervals.
  2. Lab: Build ML forecasts (LightGBM) with feature sets and produce quantile estimates; compare to baselines.
  3. Assignment: Create a forecast pipeline notebook + documentation on feature importance and production considerations.

Inventory optimization & replenishment policies

  1. Topics: classic policies (EOQ, s,S, ROP), safety stock with probabilistic lead time and demand, service level trade-offs, multi-echelon inventory basics.
  2. Lab: Implement (s,S) or ROP simulation and compute service level vs inventory trade-offs; use forecasts as input.
  3. Assignment: Recommend reorder policy for SKU/warehouse and simulate KPI outcomes (fill rate, days of inventory).

Lead‑time & supplier risk modelling; procurement analytics

  1. Topics: lead‑time prediction, supplier on‑time performance, supplier scoring, spend analytics, supplier consolidation/segmenting (Kraljic matrix), fraud/anomaly detection in procurement.
  2. Lab: Build supplier risk / lead‑time prediction model; do spend clustering and supplier segmentation; detect anomalous invoices.
  3. Assignment: Supplier risk dashboard mockup + model to alert on high‑risk suppliers.

Unstructured data: invoices, contracts, images, sensor data

  1. Topics: OCR for invoices (Tesseract, AWS Textract), NLP for contract clause extraction, classification of procurement descriptions, computer vision for warehouse (barcode/slot recognition), IoT telemetry analytics.
  2. Lab: OCR + NLP pipeline to extract PO/invoice fields and match invoices to POs; simple CV model to detect shelf/bin occupancy.
  3. Assignment: Prototype invoice extraction pipeline and error analysis.

Simulation, reinforcement learning & prescriptive analytics

  1. Topics: Discrete event simulation (SimPy), digital twins, reinforcement learning for dynamic pricing, dynamic routing, or picking policies; when to use RL vs heuristics.
  2. Lab: Simulate picking/packing process; simple RL agent (stable-baselines) to optimize picking sequence or replenishment under stochastic demand.
  3. Assignment: Compare simulation + RL policy to rule‑based baseline; show KPIs.

Deployment, governance, ethics & capstone presentations

  1. Topics: model deployment, APIs, MLOps monitoring, model drift, explainability for procurement decisions, supplier fairness and anti-competition considerations, data governance.
  2. Capstone: team presentations of end‑to‑end projects: problem, data, models/optimization, evaluation, deployment plan and ethical/governance checklist.
  3. Assessment: peer feedback + instructor scoring.

Hands‑on labs & tools

  1. Languages: Python recommended — pandas, scikit‑learn, XGBoost/LightGBM, Prophet, statsmodels, PyTorch/TensorFlow (for CV/RL/NLP).
  2. Optimization & routing: Google OR‑Tools, PuLP, cvxpy, networkx.
    Simulation & RL: SimPy, gym, stable-baselines3.
  3. OCR & NLP: Tesseract, AWS Textract/GCP Document AI (optional), Hugging Face Transformers, spaCy.
  4. Dashboards & deployment: Streamlit, Dash, FastAPI/Flask for APIs, PowerBI/Tableau for dashboards.
  5. Environments: Google Colab / Jupyter, GitHub for code, Docker for deployment.

Datasets & sources

  1. M5 Forecasting (Walmart sales forecasting)
  2. Rossmann store sales (Kaggle)
  3. Walmart or Favorita datasets (Kaggle)
  4. Instacart market basket (customer behavior)
  5. OpenStreetMap & GTFS transit feeds for routing
  6. UCI Machine Learning Repository (supply-chain related)
  7. Public procurement portals (EU/Tenders, USA Federal Procurement Data System) for spend and contract examples
  8. Simulated enterprise ERP/TMS/WMS extracts for labs (recommended to create sanitized synthetic datasets)
  9. Company/partner proprietary datasets (recommended for capstones)

Assessments & grading

  1. Weekly assignments & labs (50%): forecasting, optimization, NLP/OCR, routing.
  2. Capstone project (35%): group end‑to‑end project, code + report + presentation.
  3. Participation, quizzes & peer review (15%): in‑class exercises, short readings quizzes.
  4. Rubrics: problem framing & KPI alignment, data quality handling, model/optimization appropriateness, evaluation vs business KPIs, deployment & governance plan, reproducibility.

Capstone project ideas

  1.  SKU‑level demand forecasting and automated replenishment pilot for a regional DC.
  2. Lead‑time/arrival prediction and supplier risk scoring to prioritize expediting.
  3.  Route optimization for last‑mile deliveries including dynamic re‑routing
  4. Invoice OCR + PO‑matching pipeline to reduce manual AP exceptions.
  5. Multi‑echelon inventory optimization for a three‑tier distribution network.
  6. Anomaly detection for procurement invoices and contract noncompliance.

Suggested readings & resources

  1. Chopra & Meindl, “Supply Chain Management: Strategy, Planning and Operation” (domain basics)
  2. Hyndman & Athanasopoulos, “Forecasting: Principles and Practice” (time series)
  3. Silver, “The Signal and the Noise” (forecasting concepts)
  4. OR‑Tools documentation (routing/optimization)
  5. Papers and blogs on ML in supply chain (Gartner, McKinsey on AI in supply chain)
  6. Hugging Face tutorials (NLP)
  7. Google Cloud / AWS whitepapers on supply chain analytics
    Starter materials to share with learners
  8. GitHub repo with: syllabus, weekly notebooks, synthetic sample datasets, assignment templates, rubric.
  9.  Prebuilt Colab notebooks for forecasting, OR‑Tools routing, OCR+NLP example.
  10. Slide templates and dashboard examples (Streamlit/PBI).