arXiv subjects

Mircea-Cristian Racasan

Publications and source records attributed to Mircea-Cristian Racasan.

4 recordsLinked to original sources

arXiv2026

RAGnaroX: A Secure, Local-Hosted ChatOps Assistant Using Small Language Models

This paper introduces RAGnaroX, a resource-efficient ChatOps assistant that operates entirely on commodity hardware. Unlike existing solutions that often rely on external providers such as Azure or OpenAI, RAGnaroX offers a fully auditable, on-premise stack implemented in Rust. Its architecture integrates modular data ingestion, hybrid retrieval, and function calling, enabling flexible yet secure deployment. Our evaluation focuses on the RAG pipeline, with benchmarks conducted on the SQuAD (single-hop QA), MultiHopRAG (multi-hop QA), and MLQA (cross-lingual QA) datasets. Results show that RAGnaroX achieves competitive accuracy while maintaining strong resource efficiency, for example, reaching 0.90 context precision on single-hop questions with an average response time of 2.5 seconds per request. A replication package containing the tool, the demonstration video (https://www.youtube.com/watch? v=cDxfuEbcoM4), and all supporting materials are available at https://github.com/genius-itea/RAGnaroX.git.

cs.AR↗

arXiv2026

RAG-DIVE: A Dynamic Approach for Multi-Turn Dialogue Evaluation in Retrieval-Augmented Generation

Evaluating Retrieval-Augmented Generation (RAG) systems using static multi-turn datasets fails to capture the dynamic nature of real-world dialogues. Existing evaluation methods rely on predefined datasets, which restrict them to static, one-directional queries and limit their ability to capture the adaptive, context-dependent performance of RAG systems in interactive, multi-turn settings. Thus, we introduce the RAG-DIVE, a Dynamic Interactive Validation and Evaluation approach, that simulates user interactions with RAG systems. RAG-DIVE leverages an LLM to generate multi-turn conversations dynamically and is organized into three components. The dialogue generation stage consists of the (1) Conversation Generator, which simulates a user by creating multi-turn queries, and the (2) Conversation Validator, which filters and corrects invalid or low-quality outputs to ensure coherent conversations. The evaluation stage is handled by the (3) Conversation Evaluator, which assesses the RAG system's performance across the entire dialogue and generates both per-turn and multi-turn metrics that provide an aggregated view of system behavior. We validated RAG-DIVE through two experimental setups. First, we tested a sample RAG system, including human evaluation of dialogue quality, repeated trials to assess consistency, and an ablation study showing that RAG-DIVE detects performance changes caused by system modifications. Second, we compared RAG-DIVE with a traditional static dataset evaluation on an industrial RAG system under different configurations to verify whether both approaches reveal similar performance trends. Our findings demonstrate that RAG-DIVE facilitates dynamic, interaction-driven evaluation for multi-turn conversations, thereby advancing the assessment of RAG systems.

cs.IR↗

arXiv2024

How Industry Tackles Anomalies during Runtime: Approaches and Key Monitoring Parameters

Deviations from expected behavior during runtime, known as anomalies, have become more common due to the systems' complexity, especially for microservices. Consequently, analyzing runtime monitoring data, such as logs, traces for microservices, and metrics, is challenging due to the large volume of data collected. Developing effective rules or AI algorithms requires a deep understanding of this data to reliably detect unforeseen anomalies. This paper seeks to comprehend anomalies and current anomaly detection approaches across diverse industrial sectors. Additionally, it aims to pinpoint the parameters necessary for identifying anomalies via runtime monitoring data. Therefore, we conducted semi-structured interviews with fifteen industry participants who rely on anomaly detection during runtime. Additionally, to supplement information from the interviews, we performed a literature review focusing on anomaly detection approaches applied to industrial real-life datasets. Our paper (1) demonstrates the diversity of interpretations and examples of software anomalies during runtime and (2) explores the reasons behind choosing rule-based approaches in the industry over self-developed AI approaches. AI-based approaches have become prominent in published industry-related papers in the last three years. Furthermore, we (3) identified key monitoring parameters collected during runtime (logs, traces, and metrics) that assist practitioners in detecting anomalies during runtime without introducing bias in their anomaly detection approach due to inconclusive parameters.

cs.SE↗

arXiv2023

SoHist: A Tool for Managing Technical Debt through Retro Perspective Code Analysis

Technical debt is often the result of Short Run decisions made during code development, which can lead to long-term maintenance costs and risks. Hence, evaluating the progression of a project and understanding related code quality aspects is essential. Fortunately, the prioritization process for addressing technical debt can be expedited with code analysis tools like the established SonarQube. Unfortunately, we experienced some limitations with this tool and have had some requirements from the industry that were not yet addressed. Through this experience report and the analysis of scientific papers, this work contributes: (1) a reassessment of technical debt within the industry, (2) considers the benefits of employing SonarQube as well as its limitations when evaluating and prioritizing technical debt, (3) introduces a novel tool named SoHist which addresses these limitations and offers additional features for the assessment and prioritization of technical debt, and (4) exemplifies the usage of this tool in two industrial settings in the ITEA3 SmartDelta project.

cs.SE↗