ISW 2025

The 2nd Israeli Systems & AI Workshop

September 10, 2025

Hosted at IBM Haifa

 

Registration

ISW 2025 attendance is free of charge but requires registration ahead of time.

 

 Program

 

Keynote

Scaling LSM-Trees for the AI Age and Beyond

Niv Dayan (University of Toronto)

Abstract:
Log-Structured Merge-trees (LSM-trees) underpin large-scale storage systems such as Bigtable, DynamoDB, and RocksDB, and are foundational to ML and AI infrastructure including feature stores, recommendation engines, and vector databases. While their write-optimized design is well-suited to write-intensive workloads, scaling both read and write performance simultaneously remains a central challenge.

This talk revisits core LSM-tree design principles and presents recent advances that upend longstanding trade-offs between read and write efficiency. We discuss the use of range filters as a replacement for traditional Bloom filters and introduce techniques for optimizing filter space allocation across levels to reduce false positives and compaction costs simultaneously. These techniques unlock new regimes of performance, enabling LSM-based systems to scale more gracefully under modern demands.

Bio:
Niv Dayan is an Assistant Professor at the University of Toronto (U of T). His research focuses on designing and analyzing data structures for database and storage systems. Before joining U of T, he was a Research Scientist at Pliops and a Technical Advisor to Speedb. He received his Ph.D. from the IT University of Copenhagen and was a postdoctoral researcher at Harvard.

 

Talks

Highlight Papers – Session A

Chair: Joel Nider (Unifabrix)

• Reinforcement Learning Beyond the Lab: Real-Time Agents for the Datacenter Network
  Presented by: Gal Dalal (Nvidia Research)
  Congestion control, GBRL 
• Building ML at Scale: Robust, Fault-Tolerant, and Efficient
  Presented by: Kfir Y. Levy (Technion ECE) 
  One of the papers covered
• The Future is Fast and Affordable: Designing LLM Inferencing Platforms
  Presented by: Vita Bortnikov (IBM Research) 
  llm-d

Systems Optimized for AI – Session B

Chair: Amit Golander (Tel-Aviv University, PLiOPS)

• How to Build a Flexible Cluster of 400,000 AI accelerators
  Hezi Rahamim (Google)
• Cost-Effective LLM Inference via Disaggregated Key-Value Storage
  Eshcar Hillel (PLiOPS)
• Lord of the Memory Fabrics: Is There One Fabric Standard that can Rule them All?
  Danny Volkind (Unifabrix)
• Mavric3-based System that Accelerates Future yet-to-be-invented AI Algorithms
  Ilan Tayari (NextSilicon)

Agentic AI – Session C

Chair: Doron Chen (IBM Research)

• Building Safe and Reliable AI Agents: Dynamic Code Generation, Execution, and Permission Challenges
  Gil Vernik (IBM Research)
• LLM-Based Multi-Agent Systems for Complex Human-Centered Tasks
  Sarit Kraus (Bar-Ilan University)
• A Runtime System for Agentic AI
  Sagi Grimberg (VAST Data)

 

Venue

The IBM Haifa facility is located at the northwest edge of the University of Haifa campus.

Parking is available at IBM.