Data is often compared to oil, valuable but useless until it is refined. As global data generation keeps climbing, though, the harder problem is no longer refinement. It is plumbing, scale, and speed. Every swipe on an app, every digital advertisement served, and every cross-border financial transaction depends on invisible, highly sophisticated data pipelines that must process trillions of data points in real time without a single millisecond of lag.
Sai Sukesh Reddy Tummuri is a data engineer whose work sits inside that layer. His career spans tenures at organizations including Amazon and Meta Platforms Inc., where he has built scalable, low-latency, automated data frameworks that connect massive computing infrastructure to the strategic decisions made in the boardroom.
Engineering AdTech Infrastructure at Meta
Working in digital advertising at Meta Platforms Inc. in Menlo Park, California, Tummuri handles the volatile, high-volume side of AdTech. Within Meta’s gaming and app monetization ecosystems, he has engineered end-to-end data pipelines that process large streams of user engagement and in-app purchase attribution data.
In the AdTech space, a dropped data packet or a delayed metric can translate to millions of dollars in lost ad revenue or skewed optimization algorithms. Tummuri approached this by implementing real-time monitoring and streaming pipelines equipped with custom anomaly-detection alert rules. Those automated “pulse systems” give Meta’s stakeholders a way to flag unusual revenue spikes or drops as they happen, which supports the platform’s topline monetization reporting.
He has also worked past the limits of traditional business intelligence by developing interactive, AI-powered dashboards. Integrating automated AI agents directly into those reporting frameworks turned static charts into dynamic systems where cross-functional teams can query critical metrics and receive answers on the spot, widening self-serve access to data analytics across the organization.
Automation and Efficiency Work in FinTech at Amazon
Before his AdTech work at Meta, Tummuri spent years solving complex data and financial infrastructure puzzles within the FinTech domain at Amazon.com. Operating inside Amazon’s highly regulated TaxTech organization, he designed and maintained data platforms capable of processing vast transactional tax figures across multiple international jurisdictions, with audit readiness as a standing requirement.
One of his defining projects at Amazon was the architectural overhaul of month-end financial closures. Intercompany matching and global tax calculations had historically required intensive manual hours from finance teams worldwide. Tummuri engineered an automated matching solution that reduced manual intervention, cut operational overhead, and narrowed the openings for human error in the process.
His work also tied data engineering to fiscal efficiency. Through precise data modeling, bucketing, partitioning, and the optimization of complex Glue PySpark scripts, he reduced query execution times. A sustained focus on cloud FinOps brought down infrastructure compute costs for Amazon, a reminder that data design choices carry a direct cost consequence.
Research on Self-Healing Data Architectures
His published research includes work on Enhancing Data Pipelines with Foundation Models, which explores how large language models can automate tedious schema mapping and SQL generation to remove manual coding bottlenecks. His papers on machine learning-driven data quality monitoring and on Optimizing DAG Scheduling in Data Pipelines Using Reinforcement Learning offer blueprints for “self-healing” data architectures, systems that catch silent data corruption on their own and reschedule workflows dynamically to save cloud compute costs.
