Updates

Publications

Refereed conference publications

1. Dynamic NUMA-Aware Replication for Data Structures
   Erika Hunhoff, Zack McKevitt, Ankit Bhardwaj, Reto Achermann, Gerd Zellweger, Marcos Aguilera, Eric Keller
   40^th International Parallel & Distributed Processing Symposium (IPDPS '26)
   [ Paper ] [ Talk ] Star


2. Unleashing The Potential of Datacenter SSDs by Taming Performance Variability
   Gohar Irfan Chaudhry, Ankit Bhardwaj, Zhenyuan Ruan, Adam Belay
   23^rd USENIX Symposium on Networked Systems Design and Implementation (NSDI '26)
   [ Paper ] [ Talk ] Star


3. Checkmate: Zero Performance Overhead Model Checkpointing via Network Gradient Replication
   Ankit Bhardwaj*, Weiyang Wang*, Jeremy Carin, Adam Belay, Manya Ghobadi
   (*equal contribution)
   23^rd USENIX Symposium on Networked Systems Design and Implementation (NSDI '26)
   [ Preprint ] [ Talk ] [ Webpage ] Star


4. Auto-reconfiguration for Latency Minimization in CPU-based DNN Serving
   Ankit Bhardwaj, Amar Phanishayee, Deepak Narayanan, Ryan Stutsman
   42^nd International Conference on Machine Learning (ICML '25)
   [ Paper ] [ Poster ] [ Webpage ] Star


5. NrOS: Effective Replication and Sharing in an Operating System
   Ankit Bhardwaj, Chinmay Kulkarni, Reto Achermann, Irina Calciu, Sanidhya Kashyap, Ryan Stutsman, Amy Tai, and Gerd Zellweger
   15^th USENIX Symposium on Operating Systems Design and Implementation (OSDI '21)
   [ Paper ] [ Talk ] [ Webpage ] Star


6. Adaptive Placement for In-memory Storage Functions
   Ankit Bhardwaj, Chinmay Kulkarni, and Ryan Stutsman
   2020 USENIX Annual Technical Conference (ATC '20)
   [ Paper ] [ Talk ] Star

Refereed workshop publications

1. ObjectTier: Non-invasively Boosting Memory Tiering Performance
   Vinita Pawar, Ankit Bhardwaj, Ryan Stutsman
   8^th Workshop on Hot Topics in Cloud Computing Performance (HotCloudPerf '25)
   [ Paper ]


2. Cache-Coherent Accelerators for Persistent Memory Crash Consistency
   Ankit Bhardwaj, Todd Thornley, Vinita Pawar, Reto Achermann, Gerd Zellweger, Ryan Stutsman
   14^th ACM Workshop on Hot Topics in Storage and File Systems (HotStorage '22)
   Best Paper Award
   [ Paper ] [ Slides ] [ Talk ]


3. On the Impact of Isolation Costs on Locality-aware Cloud Scheduling
   Ankit Bhardwaj, Meghana Gupta, and Ryan Stutsman
   12^th USENIX Workshop on Hot Topics in Cloud Computing (HotCloud '20)
   [ Paper ] [ Talk ]


4. A Preliminary Performance Model for Optimizing Software Packet Processing Pipelines
   Ankit Bhardwaj, Atul Shree, Bhargav Reddy V, and Sorav Bansal
   8^th ACM SIGOPS Asia-Pacific Workshop on Systems (APSys '17)
   [ Paper ] [ Poster ] [ Slides ] [ Talk ] [ Full Paper (unpublished) ]

Research

I build low-latency, scalable systems for modern datacenters. My current work spans ML infrastructure, storage systems, and OS design.

• Checkmate enables zero-overhead model checkpointing for distributed training by replicating gradients over the network, eliminating training pauses caused by checkpoint I/O. [NSDI '26]
• Packrat is an auto-reconfiguration framework for CPU-based DNN serving that discovers optimal operator parallelism and batching strategies to minimize latency. [ICML '25]
• Sandook addresses SSD performance variability in datacenters, unlocking their full potential for latency-sensitive applications. [NSDI '26]
• NrOS is a Rust-based operating system that replicates kernel state per NUMA node and uses operation logging to synchronize replicas, scaling efficiently across multi-socket machines. [OSDI '21]
• Splinter is a multi-tenant in-memory key-value store that leverages Rust's type system to provide microsecond-scale tenant isolation while enabling runtime storage function extensibility. [ATC '20]
• Bacus applies compiler-driven optimizations to software packet processing, improving the performance and usability of DSLs such as P4. [APSys '17]
• vLAB is a system for provisioning and managing large numbers of virtual machines on commodity hardware using distributed object storage.

Teaching

• CS 151, Fall 2025 - Fundamentals of Disaggregated Systems
• CS 111, Spring 2026 - Operating Systems

Service

    Program Committee:

• OSDI '26
• ATC '25 (Heavy)
• NSDI '25
• USENIX ATC '24 (External)
• HotCloud '20 (External)