Md Shahbaz Alam

Automated Code Fix Platform using LLM · Built a Streamlit platform with AWS Bedrock LLM to monitor Glue and Lambda errors via CloudWatch. · Impact: Cut MTTR, saving €2.6M in 6 months. 200TB+ Analytics Data Pipeline Architecture · Designed a high-volume data ingestion pipeline into Redshift using EC2, Lambda, Glue (PySpark), S3 Iceberg, and Redshift RA3 nodes. · Impact: Reduced latency by 40%.

EDD Prediction Model · Built a Random Forest model using order data and distance, improving delivery date accuracy from 65% to 90%. · Impact: Reduced customer complaints and increased delivery transparency by 25%. Location-Based Product Tiles · Developed location-aware product recommendations, boosting click-to-purchase rates from 45% to 80%. · Impact: Doubled conversion rates, significantly improving sales.

Transaction & Payment Reconciliation · Automated the reconciliation process across 12+ banks and 3 key systems, handling 80K+ daily transactions to streamline validation and reduce manual errors. · Impact: Cut payment error rate from 5% to 1%, saving €148K annually. Revenue Forecasting Model · Developed 6 predictive models using historical user and order data to forecast revenue across categories over 6–12 months. · Impact: Improved planning accuracy and drove targeted strategies, increasing MRR by €10K.

Project I: Customer Segmentation via Purchasing Behaviour · Conducted EDA on retail data and applied KMeans/Hierarchical clustering to segment customers, visualising insights via Tableau for targeted marketing strategies. Tech Stack: R, SQL, Tableau Project II: Predicting Medication Non-Adherence · Identified adherence patterns using ANOVA and correlation analysis, then built classification models (Logistic Regression, Random Forest) to predict non-compliance with 80% accuracy. Tech Stack: R, SQL, Tableau