Saurabh Chhajed

Client: Leading U.S. Home Improvement Retailer

• Spearheaded architecture and delivery of a scalable Promotion Analytics Engine, driving campaign insights and analytics based on 20+ financial parameters.
• Developed robust, production-grade, DBT models based ETL pipelines combining PySpark and BigQuery for batch and near-real-time transformations.
• Integrated LightGBM regression models via BigQuery ML to enable accurate volume forecasting.
• Led a team of 10 engineers, overseeing design, architecture, and code reviews ensuring quality and scalability.
• Collaborated across various stakeholders – Product, Data Science, and Architects.

Salesforce's Unified Intelligence Platform (UIP) is an enterprise-scale internal data lake and analytics ecosystem, facilitating petabyte-scale data ingestion, exploration, transformation, and visualization.

AWS EMR | S3 | Airflow | Spark | Scala | Python | Kubernetes | Docker | Trino | Iceberg | Jupyter Notebooks

• Led the architecture and development of a metadata-driven ingestion pipeline processing petabytes of data, integrating Kafka, Spark, Scala, Trino, and Airflow for scalable batch and streaming ingestion.
• Designed GDPR-compliant data leak scanners with sampling techniques, cutting scanning costs by 40%.
• Built a high-throughput Leak Management Pipeline cleaning leaked PII data across 3000+ record types and several PBs of data.
• Engineered an advanced tokenization service securing sensitive identifiers while enabling efficient analytics.
• Developed Airflow Operators and frameworks to streamline ingestion and exploration workflows.
• Implemented system monitoring and alerting with Grafana and PagerDuty for real-time system visibility.
• Created a workload analytics dashboard using Apache Superset, optimizing Spark cluster resource utilization by ~20%.
• Led a team of 5 engineers, driving design reviews, code quality initiatives, and operational improvements.

• Architected and built a merchant recommender system using an ensemble of CatBoost, collaborative filtering, and Word2Vec models on Spark, enhancing Amex marketing personalization.
• Developed end-to-end ML pipelines: feature extraction, model training, hyperparameter tuning (distributed grid search), monitoring, and deployment to online scoring systems.
• Reduced hyperparameter tuning time by 40% through distributed Spark-based optimization.
• Designed and deployed microservices-based model serving architecture for real-time, geo-personalized merchant recommendations.
• Built Model/Feature Monitoring solutions to track GINI, PSI, and accuracy metrics, ensuring model health.
• Spearheaded development of an online offer personalization engine with Hadoop, Hive, MapRDB, and Elasticsearch, improving campaign launch speed by 50%.
• Collaborated cross-functionally with Product, Data Science, and Marketing teams using Agile/SAFe methodologies.

• Developed a multi-clustered distributed data management platform for high availability, low-latency processing.
• Built real-time Search and Analytics solutions using ELK Stack (Elasticsearch, Logstash, Kibana).
• Designed real-time order update systems using distributed caching (Gemfire).
• Evangelized code quality tools (Sonar, Jira, Crucible), improving team code health.
• Conducted a Hadoop PoC for analyzing cross-application usage patterns.
• Worked on Messaging products, providing end to end integration between many business-critical applications involving app. 2-3 million message exchanges daily.
• Exposure to Multithreading and Java performance tuning methodologies involving GC algorithms and tuning.

• Led re-architecture of a large-scale ASP/IIS application to a Java/Spring microservices framework.
• Designed and developed RESTful APIs consumed by multiple clients.
• Improved critical business process execution time by 50%, saving $30,000.
• Optimized database queries and developed complex PL/SQL procedures.