Lavdim Menxhiqi

Mr. Lavdim Menxhiqi

Faculty of Computer Science and Engineering, University for Business and Technology, Pristina, Kosovo
+1
orcid

Research Focus & Metrics

Academic Career Length

8Years (2019-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
4 July 2025 in Applied Sciences
Verified
Journal Article
Open access
PDF file
BQBesnik Qehaja
Besnik Qehaja
EHEdmond Hajrizi
Edmond Hajrizi
BHBehar Reshat Haxhismajli
Behar Reshat Haxhismajli
Lavdim MenxhiqiLavdim Menxhiqi
Lavdim Menxhiqi
Galia MarinovaGalia Marinova
Galia Marinova
EMElissa Mollakuqe
Elissa Mollakuqe
This paper presents a mathematical and statistical analysis of the security–performance trade-off in the context of the IoT Cloud architecture implemented at UBT Smart City. Through detailed modeling and real-world measurement data collected before and after the deployment of advanced security measures—such as VPN configuration, Network Security Groups (NSGs), Route Tables, and DDoS Protection—we quantify the impact of security on system performance. We propose a mathematical framework to evaluate the propagation delay of telemetry data through the system and employ queueing theory (M/M/1 model) to simulate the behavior of critical data processing services. Additionally, we perform hypothesis testing and statistical comparison to validate the significance of the observed performance changes. The results show an average delay increase of approximately 19% following the implementation of security mechanisms, highlighting the inevitable trade-off between enhanced security and operational speed. Finally, we introduce a multi-objective cost-delay function that can guide the selection of optimal security configurations by balancing latency and cost, providing a valuable tool for the future optimization of secure IoT infrastructures

1 January 2019 in IFAC-PapersOnLine
Verified
Journal Article
Closed access
EHEdmond Hajrizi
Edmond Hajrizi
BZBlerim Zylfiu
Blerim Zylfiu
Lavdim MenxhiqiLavdim Menxhiqi
Lavdim Menxhiqi

1 June 2026 in Preprints.org
Verified
Preprint
Open access
PDF file
Lavdim MenxhiqiLavdim Menxhiqi
Lavdim Menxhiqi
Galia MarinovaGalia Marinova
Galia Marinova
In a closed-inventory Large Language Model (LLM) system such as Online-CADCOM, which recommends engineering tools from a verified inventory, mention-level hallucination occurs when the model recommends tools not present in the inventory. We evaluate a three-mechanism mitigation stack consisting of database-grounded context injection, fixed vocabulary constraints, and enforced JavaScript Object Notation (JSON) output across three commercial LLM providers (OpenAI, Anthropic, Google), two model generations, and two output modes (standard and reasoning), totaling 6,912 Application Programming Interface (API) calls over 12 configurations. The hallucination rate (HR) decreases from 59–74% to 3.3–14.9% under the full architecture, with cross-provider averages stable across generations (6.8% Generation 1 (Gen1), 7.5% Generation 2 (Gen2)). A key finding is the C3 anomaly, observed under the configuration where only JSON output enforcement is active without any grounding mechanisms: JSON enforcement alone increases hallucination above the unconstrained baseline for all providers (+10.1 percentage points (pp) Gen1, +15.1 pp Gen2). We hypothesize that mandatory entity fields in the JSON schema exert slot-filling pressure, forcing models to populate tool-name slots from training-data priors when no grounding vocabulary is provided. This hypothesis requires further experimental validation. Reasoning-mode models provide no statistically significant improvement under architectural constraints. A frequency-weighted audit shows that the majority of remaining out-of-inventory mentions correspond to real engineering tools absent from the platform's inventory. Under the full architecture, 35–63% of responses still contain at least one such mention, indicating that handling unseen tools remains an open challenge for closed-inventory recommendation systems.

22 July 2026 in AI
Verified
Journal Article
Open access
PDF file
Lavdim MenxhiqiLavdim Menxhiqi
Lavdim Menxhiqi
Galia MarinovaGalia Marinova
Galia Marinova
In a closed-inventory large language model (LLM) system such as Online-CADCOM, which recommends engineering tools from a verified inventory, we measure inventory non-compliance, that is, a mention-level event in which the model recommends a tool not present in the verified inventory. We use this inventory-relative sense of hallucination throughout: an out-of-inventory mention may be a fabricated tool or a real commercial tool absent from the curated inventory, so the metric reports inventory non-compliance rather than factual fabrication. We evaluate a three-mechanism mitigation stack consisting of database-grounded context injection, fixed vocabulary constraints, and enforced JavaScript Object Notation (JSON) output across three commercial LLM providers (OpenAI, Anthropic, Google), two model generations, and two output modes (standard and reasoning), totaling 6912 Application Programming Interface (API) calls over 12 configurations. Under a recall-equalized detector adopted as the primary metric, the inventory non-compliance rate, which we denote the hallucination rate (HR) following common usage, decreases from roughly 69–80% to 4–13% under the full architecture. The cross-provider average is similar across the two generations tested (8.5% Generation 1 (Gen1), 6.9% Generation 2 (Gen2)), although per-provider directions diverge. We also examine the C3 configuration, in which only JSON output enforcement is active without grounding. A naive detector reports a large hallucination increase over the unconstrained baseline (+10.1 percentage points (pp) Gen1, +15.1 pp Gen2), but we show this gap is largely a detection-format artifact: structured JSON fields make out-of-inventory tools easy to extract, whereas the same real tools are frequently missed in free text. Under a recall-equalized detector the gap narrows to +2.6 pp (Gen1) and +4.8 pp (Gen2) and remains statistically significant only for two current-generation models, indicating a small, current-generation effect rather than a universal one. Reasoning-mode models provide no statistically significant improvement under architectural constraints. A frequency-weighted audit shows that the majority of remaining out-of-inventory mentions correspond to real engineering tools absent from the platform’s inventory. Under the full architecture, roughly half of responses (pooled Pany≈49.5%) still contain at least one such mention, indicating that handling unseen tools remains an open challenge for closed-inventory recommendation systems. Our evidence comes from a single engineering platform with four related electronic-design and power-electronics domains, so the findings characterize this setting rather than recommendation domains in general.

9 July 2024
Verified
Conference Paper
Closed access
Lavdim MenxhiqiLavdim Menxhiqi
Lavdim Menxhiqi
GMGalia Marinova
Galia Marinova
This paper explores the integration of Artificial Intelligence (AI) in Printed Circuit Board (PCB) design tools, highlighting the transformative impact AI can have on PCB design workflows. Based on the available documentation and secondary sources, nine AI-powered PCB design tools are analyzed, with a discussion of their features, strengths, and limitations. Due to access constraints, only Flux.ai was tested directly, serving as a focused case study. The case study demonstrates the capabilities of Flux.ai's AI Copilot feature, while also acknowledging its limitations. Our findings advocate for a future where AI-driven design tools enhance efficiency, precision, and user-friendliness in PCB design, setting new benchmarks for the field.

12 December 2025 in Proceedings of International Conference on Intelligent Systems and New Applications
Verified
Journal Article
Open access
PDF file
Lavdim MenxhiqiLavdim Menxhiqi
Lavdim Menxhiqi
The Online-CADCOM platform operates as a cloud-based decision support system which lets users pick Computer-Aided Design (CAD) tools for telecommunications and electronics applications. The production platform uses two Multi-Criteria Decision Analysis (MCDA) methods MAUT and PROMETHEE II to evaluate tools through binary feature criteria stored in a PostgreSQL knowledge base [3], [4]. The MCDA engine receives three additional decision analysis methods which include TOPSIS (Technique for Order Preference by Similarity to Ideal Solution) and VIKOR (VlseKriterijumska Optimizacija I Kompromisno Resenje) and COPRAS (Complex Proportional Assessment). The research module contains all five evaluation methods which process equal decision matrices to generate agreement metrics through Spearman rank correlation and top-k overlap analysis. The evaluation of three actual selection cases including PCB design tool selection and PCB calculator selection and SMPS design tool selection produced similar decision patterns. The three methods MAUT and PROMETHEE II and VIKOR create a consensus group which produces identical rankings throughout most evaluation scenarios while COPRAS follows MAUT patterns and TOPSIS produces different results when criteria coverage is limited or when criteria have negative correlations with value-based methods. The five methods produce identical rankings because their tools display different characteristics. The research adds three main contributions to the field: (1) The Online-CADCOM engine now supports TOPSIS and VIKOR and COPRAS as additional decision analysis methods. (2) The research evaluates five MCDA methods through actual tool passport data from three engineering fields. The research establishes essential guidelines which engineers need to select proper MCDA methods for their tool selection work.