Congratulations to our amazing undergraduate student, Regan McDonald, and the team on their upcoming paper at the SIGCOMM NAIC '26 Workshop!
Murayyiam did a fantastic job presenting SpliDT at P4 Dev Day!
🗓️ TOMORROW: March 19 at 11 am ET/4 pm CET ✨ P4 Developer Day || SpliDT: Partitioned Decision Trees for Scalable Stateful Inference at Line Rate with Murayyiam Parvez 🎟️ Register here:… | P4 Langua
🗓️ TOMORROW: March 19 at 11 am ET/4 pm CET ✨ P4 Developer Day || SpliDT: Partitioned Decision Trees for Scalable Stateful Inference at Line Rate with Murayyiam Parvez 🎟️ Register here: https://lnkd.in/dMZWEVBj Abstract: Machine learning is increasingly used in programmable data planes, such as switches and smartNICs, to enable real-time traffic analysis and security monitoring at line rate. Decision trees (DTs) are particularly well-suited for these tasks due to their interpretability and co
Woohoo! The NextGArch Lab is on a roll—celebrating its second PhD graduate stepping into the real world. Huge congratulations to the outstanding Dr. Annus Zulfiqar!
Our own, Venkat Kunaparaju, is joining NVIDIA to work on all things networking (for AI and Cloud Gaming). Congratualtion, Venkat!
Huge congratulations, Venkat Kunaparaju! 🎉 It has been a pleasure having you as part of our NextGArch Lab since 2023 ... quickly standing out for your initiative, technical depth, and ability to… |
Huge congratulations, Venkat Kunaparaju! 🎉 It has been a pleasure having you as part of our NextGArch Lab since 2023 ... quickly standing out for your initiative, technical depth, and ability to turn ideas into real systems work. Your contributions to our GigaFlow work on scalable fast paths for Open vSwitch and SmartNICs have been outstanding—presenting it at TechCon and then taking it through HotCHIPs and ASPLOS! 🙏 For an undergraduate student to contribute at this level—across architect
Prof. Shahbaz recognized as a Michigan Housing Honored Instructor. Congratulations!
Michigan Housing
Michigan Housing Honored Instructors
Since 2018, Michigan Housing has provided residential students an opportunity to honor the instructors who make a positive impact on their collegiate journey at the University of Michigan. Michigan Housing is excited to continue to celebrate the incredible faculty and instructors that inspire our students each and every day.
Ertza will be presenting our recent work on OptiNIC at the OCP's Time Appliances Project (TAP).
For our next OCPTAP session, we have Ertza Warraich, systems and networking researcher and recent Ph.D. graduate from Purdue University. Ertza will present OptiNIC, a domain-specific RDMA transport… |
For our next OCPTAP session, we have Ertza Warraich, systems and networking researcher and recent Ph.D. graduate from Purdue University. Ertza will present OptiNIC, a domain-specific RDMA transport designed for large-scale distributed machine learning. His talk explores how relaxing traditional reliability and in-order delivery guarantees can dramatically reduce tail latency and improve throughput across multi-GPU, high-speed interconnects. The session will cover: • Why strict RDMA semantics be
SpliDT accepted to NSDI '26. Congratulations, Murayyiam Parvez, Annus Zulfiqar, and the team!
SPLIDT Accepted to NSDI2026: Scalable Stateful Inference at Line Rate | Muhammad Shahbaz posted on the topic | LinkedIn
🚨 Big and humbling news! Our paper SPLIDT: Partitioned Decision Trees for Scalable Stateful Inference at Line Rate has been accepted to #NSDI2026! 🎉 In-network ML has long been caught between a rock and a hard place—accuracy or scalability. SPLIDT says: why not both? SPLIDT reimagines how decision trees operate in programmable data planes by: • ✂️ Partitioning trees into subtrees with their own stateful features, • 🔁 Recirculating packets to reuse registers and match-action tables (MATs) ac
Woo hoo! NextGArch Lab proudly celebrates its very first PhD graduate—congratulations to the one and only Dr. Ertza Warraich!
Towards Network-Efficient Cross-Regional Inference via Learned Activation Compression
Regan McDonald, Marilyn Rego, Ertza Warraich, Annus Zulfiqar, Muhammad Shahbaz
mrLLM: Fast Multi-Region LLM Inference using Learned Adaptors
Marilyn Rego, Maxwell Kumbong, Hermann Kumbong, Ertza Warraich, Muhammad Shahbaz
SpliDT: Partitioned Decision Trees for Scalable Stateful Inference at Line Rate
Murayyiam Parvez*, Annus Zulfiqar*, Roman Beltiukov, Shir Landau Feibish, Walter Willinger, Arpit Gupta, Muhammad Shahbaz (*co-primary)
33:26
YouTube
SpliDT: Partitioned Decision Trees for Scalable Stateful Inference at Line Rate || P4 Developer Day
Machine learning is increasingly used in programmable data planes, such as switches and smartNICs, to enable real-time traffic analysis and security monitoring at line rate. Decision trees (DTs) are particularly well-suited for these tasks due to their interpretability and compatibility with the Reconfigurable Match-Action Table (RMT) architecture. However, current DT implementations require collecting all features upfront, which limits scalability and accuracy due to constrained data plane reso
Reimagining RDMA Through the Lens of ML
Ertza Warraich, Ali Imran, Annus Zulfiqar, Shay Vargaftik, Sonia Fahmy, Muhammad Shahbaz
NetSparse: In-Network Acceleration of Distributed Sparse Kernels
Gerasimos Gerogiannis, Charles Block, Dimitrios Merkouriadis, Annus Zulfiqar, Filippos Tofalos, Muhammad Shahbaz, Josep Torrellas
SpliDT: Partitioned Decision Trees for Scalable Stateful ML Inference at Line Rate
Marilyn Rego, Murayyiam Parvez, Annus Zulfiqar, Roman Beltiukov, Shir Landau Feibish, Walter Willinger, Arpit Gupta, Muhammad Shahbaz
O'MINE: A Novel Collaborative DDoS Detection Mechanism for Programmable Data-Planes
Enkeleda Bardhi, Chenxing Ji, Ali Imran, Muhammad Shahbaz, Riccardo Lazzeretti, Mauro Conti, Fernando Kuipers
HardHarvest: Hardware-Supported Core Harvesting for Microservices
Jovan Stojkovic, Chunao Liu, Muhammad Shahbaz, Josep Torrellas