CV

Click here to view the PDF version

Education

• Ph.D. in Computer Science, Syracuse University, Expected July 2026
• M.S. in Computer Engineering, Stevens Institute of Technology, 2019
• B.E. in Biomedical Engineering, University of Mumbai, 2017

Work Experience

• Research Assistant, S4 Lab, Syracuse University — August 2020 – Present
  • Researching storage and I/O systems for large-scale ML workloads under Prof. Bryan Kim, supported by NSF grant CPR for Flash-Based Storage Systems.
  • Built Seneca (FAST ‘26): a data preprocessing system for ML training that decouples preparation from training epochs, eliminating I/O bottlenecks and improving end-to-end GPU utilization.
  • Built Cicero (EuroMLSys ‘26): a cost-efficient training and checkpointing framework for large models on preemptible cloud VMs.
  • Developed an open-source remote monitoring tool for SSDs to study performance and reliability under production workloads.
  • Studied key-value stores (LevelDB, RocksDB) at depth, contributing to advances in background I/O scheduling and storage-level indexing.
• Research Intern, Samsung Semiconductor — San Jose, CA — May–August 2022
  • Designed and prototyped a task management system for computational storage devices, resulting in US Patent 12,411,630 (granted September 2025).
  • Worked with the Advanced Memory Lab on near-storage compute architectures.
• Research Intern, Samsung Semiconductor — San Jose, CA — May–August 2021
  • Designed a two-level indexing architecture for key-value persistent storage devices, resulting in US Patent 11,954,345 (granted April 2024).
  • Prototyped the design on Samsung’s key-value SSD platform and evaluated performance against baseline KV-SSD implementations.
• Software & Data Engineer, Practo Technologies — June 2019 – September 2020
  • Owned the user behavior and events ingestion pipeline — the backbone of data analytics, ML, and recommendations infrastructure at India’s largest telemedicine platform.
  • Deployed and maintained Apache Airflow on Kubernetes; wrote in-house libraries and deployment automation used across the data engineering team.
  • Brought deep visibility into the production Kubernetes cluster using Prometheus and Grafana.
  • Managed a multi-terabyte production database cluster; built data visualization tools for business analysts.
• Software Engineering Intern, Delos Living — New York, NY — May–December 2018
  • Designed and built a cloud-native IoT platform for Delos Labs’ indoor wellness research program.
  • Developed a data pipeline ingesting from a heterogeneous set of wearables and environmental sensors (~1,000 data points/hour), powering ML models for human stress and sleep research.

Skills

• Research Areas: Storage Systems, ML Infrastructure & I/O, Key-Value Stores, Distributed Systems, Flash-Based SSDs, Operating Systems
• Languages: C, C++, Python, Java
• Storage & Databases: LevelDB, RocksDB, NVMe/KV-SSD, Unix File Systems, RAID
• Infrastructure: Kubernetes, Docker, Prometheus, Grafana, Nginx
• Data Engineering: Apache Airflow, Apache Kafka, Apache Hadoop, Redis
• Tools: eBPF/BCC, fio, blktrace, Git

Publications

Home Automation Using Wi-Fi and Sensor Networks

Bhatt Parth, Desai Omkar. (2017). "Home Automation Using Wi-Fi and Sensor Networks" DJSCE Technofocus. ISSN:2321-0532.

A principled approach for selecting block I/O traces

Omkar Desai, Seungmin Shin, Eunji Lee, and Bryan S. Kim. 2022. A principled approach for selecting block I/O traces. In Proceedings of the 14th ACM Workshop on Hot Topics in Storage and File Systems (HotStorage '22). Association for Computing Machinery, New York, NY, USA, 52–58. https://doi.org/10.1145/3538643.3539754

Leveraging keys in key-value SSD for production workloads

Manoj Pravakar Saha, Omkar Desai, Bryan S. Kim, and Janki Bhimani. 2023. Leveraging Keys In Key-Value SSD for Production Workloads. In Proceedings of the 32nd International Symposium on High-Performance Parallel and Distributed Computing (HPDC '23). Association for Computing Machinery, New York, NY, USA, 327-328. https://doi.org/10.1145/3588195.3595949

Two-level indexing for key-value persistent storage device

Omkar Desai, Changho Choi, and Yangwook Kang. 2024. Two-level indexing for key-value persistent storage device. U.S. Patent 11,954,345, filed February 9, 2022, granted April 9, 2024. Assigned to Samsung Electronics Co., Ltd.

System and method for managing tasks in a storage device

Omkar Desai, Shuyi Pei, Jing Yang, and Rekha Pitchumani. 2025. System and method for managing tasks in a storage device. U.S. Patent 12,411,630, filed July 17, 2023, granted September 9, 2025. Assigned to Samsung Electronics Co., Ltd.

Preparation meets opportunity: enhancing data preprocessing for ML training with Seneca

Omkar Desai, Ziyang Jiao, Shuyi Pei, Janki Bhimani, and Bryan S. Kim. 2026. Preparation Meets Opportunity: Enhancing Data Preprocessing for ML Training With Seneca. In 24th USENIX Conference on File and Storage Technologies (FAST '26), Santa Clara, CA, USA, 221-238.

Cost-efficient training and checkpointing for large models on preemptible cloud VMs

Omkar Desai, Shuyi Pei, Janki Bhimani, and Bryan S. Kim. 2026. Cost-Efficient Training and Checkpointing for Large Models on Preemptible Cloud VMs. In Proceedings of the 6th Workshop on Machine Learning and Systems (EuroMLSys '26).

Service and Leadership

• Reviewer, ACM Transactions on Storage (TOS)
• Artifact Evaluation Reviewer, USENIX NSDI 2026
• Artifact Evaluation Reviewer, USENIX FAST 2025
• Contributor to Apache Airflow — see pull requests
• Contributor to iovisor/bcc — see pull requests