Negin Lava

About Negin Lava

I have 5 years of experience building ETL pipelines for batch and real-time data using PySpark, Airflow, Python and SQL integrating multiple database tools such as Hive, BigQuery, and Cassandra.I deploy ML models into production and develop end-to-end feature pipelines for production ML systems.Additionally, I build transactional datasets, create metric dashboards, and perform ad hoc analyses to support on-call and business needs.

Experience

1. Machine Learning Engineer

   Walmart Global Tech

   Mar 2023 — Present · Sunnyvale, CA, US

   Built real-time and batch data pipelines in Python and PySpark using Cassandra, Hive and BigQuery to process 250K e-commerce transactions per day for a credit card risk scoring model. Designed low latency feature engineering pipelines and data ingestion workflows for online inference and offline model training, reducing e-commerce fraud by 76%(~$67M) while lowering false positives by 75% and rejection rates by 5.3%. Analyzed existing return abuse processes and blocklist workflows to identify inefficiencies in the return blocklist system. Designed a ranking based scoring approach and built a batch ETL pipeline using PySpark, Hive, BigQuery, and Airflow to compute and update member risk scores through nightly batch processing. Developed Looker Studio dashboards to provide transparency into blocking decisions, reducing false positives by 96% and significantly lowering manual review workload. Redesigned ETL pipelines from full overwrites to incremental batch processing, implemented load throttling and optimized table partitioning, reducing Cassandra compute costs, cutting job runtimes from hours to minutes, and improving read latency by ~10ms for real-time inference. Adapted existing ETL pipelines to process new fields in transaction payloads and Kafka events, updating XML to JSON transformations; modified Cassandra, Hive, and BigQuery schemas to support the updated data structures, and implemented pipeline changes to maintain seamless order processing in production and support downstream analytics and workflows. Built automated monitoring and alerting for ETL and ML pipelines using PySpark, Hive, Airflow, and BigQuery on GCP Dataproc clusters, for faster incident response and data accuracy; built weekly and monthly email alerts to track model performance metrics such as chargebacks and rejection rates, detecting performance degradation and feature drift before impacting downstream analytics.

Education

• Bowling Green State University

  Master's degree, Applied statistics: Business Analytics

  2020 — 2022

• Amirkabir University of Technology - Tehran Polytechnic

  Bachelor's degree, Applied Mathematics and Computer Science

• The University of Toledo

  Master's degree, Applied Mathematics

  2019 — 2020