As enterprises increasingly rely on artificial intelligence, cloud platforms and large-scale data systems, the ability to make those technologies faster, more scalable and economically efficient has become a critical engineering challenge. Srihari Babu Godleti, a Data and AI Engineering Leader at Roku Inc., is contributing to this evolving field through enterprise technology leadership and a growing body of published research focused on cloud analytics, artificial intelligence and distributed data systems.
Godleti currently works on enterprise data and AI platforms supporting functions including finance, advertising, subscriptions and human resources. His work has involved modernizing integrations across platforms such as NetSuite, Anaplan, Workday and Salesforce while leveraging AWS, Python, PySpark, Spark/EMR, Airflow, Trino, Athena, Aurora, MySQL and Snowflake. According to his professional profile, these initiatives have included generative AI applications, LLM-powered assistants, MCP services, authentication frameworks and evaluation systems, contributing to approximately 40% improved data availability and reducing one critical processing workflow from more than 20 hours to approximately 20 minutes.
His industry experience spans nearly two decades across data engineering, software development, cloud architecture and analytics. Prior to his current leadership responsibilities, Godleti held data engineering and technology roles with organizations including Amazon Web Services and Nike. His experience has included designing Hadoop applications on AWS, developing scalable analytics solutions and working with large datasets for data science teams.
A significant component of Godleti’s recent work has been his research into how artificial intelligence can improve the management of complex cloud analytics environments. In his 2026 research article, “LLM-Guided Cross-Platform Optimization of Cloud Analytics Workloads,” published in the International Journal of Computational and Experimental Science and Engineering, he introduced LLM-TradeOpt, a framework designed to reason across workload characteristics, system configurations and execution histories rather than relying solely on static optimization rules.
The study evaluated workloads across Amazon EMR, Apache Spark on Kubernetes and Snowflake using CloudSuite v4.0. Its results showed improvements of up to 18.7% in latency, 22.4% in throughput and 15.3% in cost compared with established baseline approaches. The research also found that the framework reduced latency variance and improved resource utilization, suggesting that AI-assisted optimization can address not only raw performance but also predictability and operational efficiency.
The research further examined how optimization priorities can change depending on an organization’s objectives. Performance-focused configurations produced lower latency and higher throughput, while cost-aware configurations reduced expenditure while maintaining acceptable performance. This approach reflects a broader shift toward treating cloud optimization as a multi-objective engineering problem rather than optimizing a single technical metric.
Godleti’s research portfolio also addresses practical challenges faced by engineers working with distributed data systems. His article “Taming Spark Data Skew with Practical Solutions,” published in the Journal of Computer Science and Technology Studies, examines one of the persistent performance problems in Apache Spark: uneven distribution of data across processing partitions. The work presents three approaches—repartitioning, key salting and broadcast joins—and explains how engineers can select the appropriate technique according to the characteristics and severity of the workload.
In another publication, “Leveraging SonarQube and Snowflake for Advanced ETL Solutions,” published in the European Journal of Computer Science and Information Technology, Godleti examined the combination of automated code-quality analysis with Snowflake’s cloud data architecture. The research explores how static analysis, memory-management practices and independently scalable compute and storage can contribute to more reliable and scalable ETL pipelines.
His research has also explored the intersection of artificial intelligence and television advertising. In “The Convergence of AI and Television: Transforming Ad Monetization in the Digital Era,” Godleti examines how computer vision, natural language processing, predictive analytics and other AI technologies can enable interactive and shoppable television experiences, creating direct pathways between viewer engagement and purchasing.
Taken together, Godleti’s professional and research work reflects an emerging model of technology leadership in which enterprise engineering and applied research increasingly inform one another. His focus spans the infrastructure required to process large-scale data, the AI systems built on top of that infrastructure, and the optimization techniques needed to make those systems more efficient.
As organizations continue moving toward AI-enabled products and cloud-native data platforms, engineers capable of connecting research concepts with production-scale technology are becoming increasingly important. Godleti’s work illustrates how advances in LLMs, distributed computing and cloud data architecture can be translated into practical approaches for improving the performance, scalability and efficiency of modern enterprise systems.
Disclaimer: This article is for general informational purposes only. Information about Srihari Babu Godleti, his professional experience, research, publications, technologies, results, and achievements is based on available sources and may change over time. Research findings, performance figures, and professional claims should be independently verified with the relevant publications or sources. No specific technical, business, or performance outcome is guaranteed.