Arup Sarker is a Ph.D. student in Computer Science specializing in Network Systems Science and Advanced Computing. His primary area of focus involves high-performance computing applied to distributed machine learning models and Big Data systems. In 2013, he obtained his Master of Science degree from the University of Dhaka, specializing in the reconstruction of gene regulatory networks using a hybrid evolutionary model. His professional interests encompass HPC frameworks, the scalability of deep learning models, and the optimization of machine learning frameworks from a systems perspective. Presently, he is actively involved in the development of a high-performance framework called Cylon, specifically designed for heterogeneous systems. 

Prior to joining UVA, Arup Sarker accumulated over 12 years of experience in the software industry, including more than 10 years spent at Samsung Research. Throughout his career, he has gained extensive expertise in developing micro-edition embedded systems for mobile and wearable platforms. Notably, he played a crucial role in the development of a highly sensitive protocol stack for mission-critical services defined by 3GPP, capable of securely handling large-scale data traffic.

Current Project

Traditional large language model (LLM) inference and retrieval-augmented generation (RAG) pipelines face critical scalability and reliability limitations. RAG systems struggle with fragmented data pipelines and I/O-heavy vector operations, while LLM inference remains non-deterministic and context-dependent, undermining reproducibility. Emerging agentic frameworks such as LangChain and LangGraph extend LLM capabilities through orchestration and tool invocation, yet they inherit deep architectural fragilities—inefficient data handling, dynamic execution uncertainty, and limited transparency for quality assurance. To address these issues, we propose Agentic DRC, a unified distributed runtime integrating Cylon’s high-performance data processing with RADICAL-Pilot’s deterministic resource management. The framework introduces a Deep RC Bridge for zero-copy data transfer between distributed dataframes and tensors, leveraging Apache Arrow and high-speed interconnects (GLOO/NCCL/UCX). This architecture eliminates serialization bottlenecks and ensures reproducible, high-throughput execution. 

Publications

• Arup Kumar Sarker, Aymen Alsaadi, Alexander James Halpern, Prabhath Tangella, Mikhail Titov, Gregor von Laszewski, Shantenu Jha, and Geoffrey Fox. Deep RC: A scalable data engineering and deep learning pipeline. Accepted on Job Scheduling Strategies for Parallel Processing, 2026

• Arup Kumar Sarker, Aymen Alsaadi, Niranda Perera, Mills Staylor, Gregor von Laszewski, Matteo Turilli, Ozgur Ozan Kilic, Mikhail Titov, Andre Merzky, Shantenu Jha, and Geoffrey Fox. Radical-cylon: A heterogeneous data pipeline for scientific computing. In Dalibor Klus´aˇcek, Julita Corbal´an, and Gonzalo P. Rodrigo, editors, Job Scheduling Strategies for Parallel Processing, pages 84–102, Cham, 2025. Springer Nature Switzerland

• Arup Kumar Sarker, Niranda Perera, Kaiying Shan, Alex Fetea, Supun Kamburugamuve, Thejaka Amila Kanewala, Chathura Widanage, Mills Staylor, Tianle Zhong, Vibhatha Abeykoon, et al. Supercharging distributed computing environments for high-performance data engineering. Frontiers in High Performance Computing, 2:1384619, 2024

• Arup Kumar Sarker, Aymen Alsaadi, Niranda Perera, Mills Staylor, Gregor von Laszewski, Matteo Turilli, Ozgur Ozan Kilic, Mikhail Titov, Andr´e Merzky, Shantenu Jha, et al. Design and implementation of an analysis pipeline for heterogeneous data. CoRR, 2024

• Niranda Perera, Arup Kumar Sarker, Mills Staylor, Gregor von Laszewski, Kaiying Shan, Supun Kamburugamuve, Chathura Widanage, Vibhatha Abeykoon, Thejaka Amila Kanewela, and Geoffrey Fox. In-depth analysis on parallel processing patterns for high-performance dataframes. Future Generation Computer Systems, 2023

• Sakib Reza, Arup Kumar Sarker, and Redwanul Haque Sourave. Electronic device for image processing and operating method of the same, January 12 2023. US Patent App. 17/832,164

• Arup Kumar Sarker, Md Khairul Islam, Yuan Tian, and Geoffrey Fox. Mvam: Multi-variant attacks on memory for iot trust computing. In Proceedings of Cyber-Physical Systems and Internet of Things Week 2023, pages 13–18. 2023

• Kaiying Shan, Arup Kumar Sarker, Niranda Perera, Damitha Lenadora, Tianle Zhong, Supun Kamburugamuve, Thejaka Amila Kanewela, Chathura Widanage, and Geoffrey Fox. Hybrid cloud and hpc approach to high-performance dataframes. In 2022 IEEE International Conference on Big Data (Big Data), pages 2728–2736. IEEE, 2022

• Arup Kumar Sarker and Felix Xiaozhu Lin. Incremental perception on real time 3d data. In Proceedings of the 23rd Annual International Workshop on Mobile Computing Systems and Applications, pages 68–73, 2022