At INFOR, I developed demand forecasting models that contributed to 15–35% improvements in forecasting accuracy through feature engineering, model tuning, and validation. I evaluated forecasts against actual demand using RMSE and MAPE.
I also developed pricing-optimization models and recommendation logic that contributed to 20–50% higher profit margins on specific projects or product subsets. I validated recommended prices against actual revenue and margins on test data.
At INFOR, I designed scalable data and ML pipelines supporting 1M+ SKUs, using Infor platforms and Python workflows with Pandas, NumPy, scikit-learn, XGBoost, and LightGBM. I collaborated with engineering and product teams to integrate models and pipelines into production systems.
At LIFAT Laboratory, I designed a real-time data-processing and recommendation workflow using Apache Kafka, Apache NiFi, and Apache Spark. I developed Word2Vec embeddings and NLP-based recommendation models integrated with Neo4j.
