# PROJECT 1 :HVAC Demand Forecasting System
Business Domain: HVAC Manufacturing and Supply Chain
Objective: Predicted monthly HVAC demand with 80% accuracy to optimize production and inventory management.
Achievements: Built a forecasting model that achieved 90% accuracy in predicting HVAC demand for the next 20 months, reducing inventory costs by 30%.
Challenges Resolved: Addressed seasonal demand variations and sparse historical data by implementing advanced feature engineering and model tuning.
Skills Learned: Time series forecasting, feature extraction, seasonal analysis, and production planning optimization.
Algorithms Used: SARIMA, Prophet, XGBoost, TimeGPT and LSTM for time series analysis.
# PROJECT 2 : Anomaly Detection in Financial Transactions
Business Domain: Semiconductor Manufacturing – Finance and Operations
Objective: Identified anomalies in 100,000+ financial and operational records to enhance transaction accuracy and detect fraud.
Achievements: Automated anomaly detection with 75% accuracy, reduced manual efforts by 60%.
Challenges Resolved: Addressed data imbalance, high dimensionality, and noise using advanced preprocessing and dimensionality reduction techniques.
Skills Learned: Unsupervised learning, financial data preprocessing, feature engineering, scalable pipeline development, and dashboard integration.
Algorithms Used: Isolation Forest, Autoencoders, DBSCAN, PCA, and LOF.
Deployment: Delivered a live system integrated with business intelligence tools for real-time anomaly detection and reporting.