Reza Soleymanifar

• Reduced cyber threat response time by 70% by developing an agentic AI system that analyzed petabyte-scale security telemetry and automated threat containment.
• Led a team of five junior engineers to ship a security telemetry visualization platform into production.
• Enabled enterprise-wide adoption of agentic AI workflows across 500+ security operations personnel.

• Delivered $3.7M in annual warranty-liability savings by predicting electric vehicle battery longevity early from manufacturing data.
• Led a team of three data scientists to ship the battery life prediction model into production.
• Deployed ML models as FastAPI microservices on Kubernetes with Docker-based CI/CD and Airflow orchestration for real-time factory execution.
• Built ETL pipelines processing 5M+ cells per year for model training and inference.

• Built automated ETL workflows in SQL to integrate multi-source budgeting data, reducing manual effort by 80%.
• Architected normalized relational schemas for project budgeting, improving governed financial reporting pipelines.

• Improved product recommendations by 40% using large language models and customer clickstream data.
• Engineered an ETL pipeline processing 12B+ clickstream events into Google BigQuery for downstream modeling.
• Fine-tuned state-of-the-art LLMs (Falcon 7B, LLaMA 2 7B, OPT) on GPUs, tailoring them to retail use cases.
• Translated AI research papers into a production-ready recommendation system aligned with business needs.

• Implemented a computer vision safety system detecting workers in danger zones and raising real-time alarms.
• Fine-tuned multiple object detection models (YOLO, EfficientDet, R-CNN) on imagery collected from food plant cameras.
• Productionized the model as an AWS SageMaker endpoint, providing real-time access for engineering teams.

Research: designed a novel ML algorithm that navigates swarms of drones in real time to provide network access to terrestrial users.