arXiv subjects

Burcu Kulahcioglu Ozkan

Publications and source records attributed to Burcu Kulahcioglu Ozkan.

6 recordsLinked to original sources

arXiv2026

Tracking State Footprints: How Agents Can Transact

Can AI agents transact? We argue that they must: as multi-agent systems (MASs) increasingly write code, deploy infrastructure, modify databases, and call web services, lost updates or stale reads can be catastrophic. Although MASs increasingly execute plans in parallel, current orchestrators do not track the state that agents read and write. As a result, concurrency anomalies manifest even in simple coding tasks. We frame MAS coordination as a data management problem and propose to describe agents by their state footprint: the state they read and write across their own local context and state, as well as the state of the orchestrator and external systems. We posit that MASs require guarantees similar to those of databases, but providing them raises new challenges and opportunities: unlike database transactions, agents do not read from a fixed schema or an isolated snapshot, and cannot be replayed deterministically upon failure. They can, however, resolve conflicts semantically instead of aborting, enabling new forms of concurrency control and conflict resolution. Towards agents that can transact, we outline a vision for next-generation agent orchestrators and transactional interfaces for external systems to participate in agentic transactions.

cs.DB↗

arXiv2026

Observability and Fault Injection for LLM-Based Multi-Agent Systems in Software Engineering

Large Language Model-based multi-agent systems are increasingly explored for software engineering tasks, but they remain difficult to inspect, debug, and evaluate under controlled failures. We present llmmas-otel, a lightweight and framework-agnostic tool that combines OpenTelemetry-based distributed tracing with fault injection for LLM-based multi-agent systems in software engineering workflows. The tool instruments agent executions with trace-aligned telemetry across workflow phases, agent steps, inter-agent communication, tool calls, and LLM invocations, and supports targeted fault injection at selected interaction points. This makes it possible to compare baseline and faulty executions in a reproducible way and inspect the effects through aligned traces and run artifacts. We describe the motivation, architecture, implementation, current capabilities, and initial validation of the tool on a minimal demo workflow and a real LLM-based multi-agent system for software development.

cs.SE↗

arXiv2026

TriCEGAR: A Trace-Driven Abstraction Mechanism for Agentic AI

Agentic AI systems act through tools and evolve their behavior over long, stochastic interaction traces. This setting complicates assurance, because behavior depends on nondeterministic environments and probabilistic model outputs. Prior work introduced runtime verification for agentic AI via Dynamic Probabilistic Assurance (DPA), learning an MDP online and model checking quantitative properties. A key limitation is that developers must manually define the state abstraction, which couples verification to application-specific heuristics and increases adoption friction. This paper proposes TriCEGAR, a trace-driven abstraction mechanism that automates state construction from execution logs and supports online construction of an agent behavioral MDP. TriCEGAR represents abstractions as predicate trees learned from traces and refined using counterexamples. We describe a framework-native implementation that (i) captures typed agent lifecycle events, (ii) builds abstractions from traces, (iii) constructs an MDP, and (iv) performs probabilistic model checking to compute bounds such as Pmax(success) and Pmin(failure). We also show how run likelihoods enable anomaly detection as a guardrailing signal.

cs.AI↗

arXiv2025

Balancing Fairness and Performance in Multi-User Spark Workloads with Dynamic Scheduling (extended version)

Apache Spark is a widely adopted framework for large-scale data processing. However, in industrial analytics environments, Spark's built-in schedulers, such as FIFO and fair scheduling, struggle to maintain both user-level fairness and low mean response time, particularly in long-running shared applications. Existing solutions typically focus on job-level fairness which unintentionally favors users who submit more jobs. Although Spark offers a built-in fair scheduler, it lacks adaptability to dynamic user workloads and may degrade overall job performance. We present the User Weighted Fair Queuing (UWFQ) scheduler, designed to minimize job response times while ensuring equitable resource distribution across users and their respective jobs. UWFQ simulates a virtual fair queuing system and schedules jobs based on their estimated finish times under a bounded fairness model. To further address task skew and reduce priority inversions, which are common in Spark workloads, we introduce runtime partitioning, a method that dynamically refines task granularity based on expected runtime. We implement UWFQ within the Spark framework and evaluate its performance using multi-user synthetic workloads and Google cluster traces. We show that UWFQ reduces the average response time of small jobs by up to 74% compared to existing built-in Spark schedulers and to state-of-the-art fair scheduling algorithms.

cs.DC↗

arXiv2025

Model-guided Fuzzing of Distributed Systems

We present a coverage-guided testing algorithm for distributed systems implementations. Our main innovation is the use of an abstract formal model of the system that is used to define coverage. Such abstract models are frequently developed in early phases of protocol design and verification but are infrequently used at testing time. We show that guiding random test generation using model coverage can be effective in covering interesting points in the implementation state space. We have implemented a fuzzer for distributed system implementations and abstract models written in TLA+. Our algorithm shows better coverage over purely random exploration as well as random exploration guided by different notions of scheduler coverage and mutation. In particular, we show consistently higher coverage and detect bugs faster on implementations of distributed consensus protocols such as those in Etcd-raft and RedisRaft. Moreover, we discovered 13 previously unknown bugs in their implementations, four of which could only be detected by model-guided fuzzing.

cs.SE↗

arXiv2023

Liveness Checking of the HotStuff Protocol Family

Byzantine consensus protocols aim at maintaining safety guarantees under any network synchrony model and at providing liveness in partially or fully synchronous networks. However, several Byzantine consensus protocols have been shown to violate liveness properties under certain scenarios. Existing testing methods for checking the liveness of consensus protocols check for time-bounded liveness violations, which generate a large number of false positives. In this work, for the first time, we check the liveness of Byzantine consensus protocols using the temperature and lasso detection methods, which require the definition of ad-hoc system state abstractions. We focus on the HotStuff protocol family that has been recently developed for blockchain consensus. In this family, the HotStuff protocol is both safe and live under the partial synchrony assumption, while the 2-Phase Hotstuff and Sync HotStuff protocols are known to violate liveness in subtle fault scenarios. We implemented our liveness checking methods on top of the Twins automated unit test generator to test the HotStuff protocol family. Our results indicate that our methods successfully detect all known liveness violations and produce fewer false positives than the traditional time-bounded liveness checks.

cs.DC↗