American Research Journal of Computer Science and Information Technology

Venkata Surendra Reddy

Surging rates of machine learning deployment in real-world applications have highlighted the need for improved systems built to support large-scale, dependable and trustworthy AI solutions. Extending DevOps practices, MLOps focuses on ML lifecycle management to address the demands of handling data versions, re-training models, managing governance and delivering ML services continually. This article thoroughly examines how to use MLOps to deploy, monitor and develop machine learning models in many different work settings. We observe through history how MLOps progressed from managing models manually to the present use of automated and professional systems. key elements such as data engineering, model training using automation, validation workflow, the process to deploy the model and post-deployment monitoring are carefully looked at. This domain specializes in issues that distinguish ML systems from traditional software systems. It means a change in the inputs, change in the data and the need for frequent updates to the model. We discuss how to manage model governance effectively, paying attention to checking versions and ethical approval, as well as the importance of automating infrastructure for complex ML tasks. It also focuses on future changes in MLOps such as adopting foundation models, using AutoML for better pipeline design and focusing on edge deployment strategies. It brings together industrial processes, academic findings and case study reviews to guide researchers and practitioners in making machine learning useful for many users. The findings underscore MLOps’ pivotal role in enabling sustainable, trustworthy, and agile AI systems across industries.

Danylo Sereda

With the increasing complexity and scalability of modern applications, choosing the optimal API architectural style has become a key task for developers. REST and GraphQL are two popular approaches, each with its own unique features, advantages, and limitations. This article is devoted to comparing these two approaches in order to facilitate the decision-making process for developers and companies. We analyze in detail the differences between REST and GraphQL, highlighting their strengths and weaknesses in different usage contexts. Special attention is paid to how these differences can affect productivity, development time, and end user satisfaction. Understanding these aspects will allow you to choose the approach that best meets the requirements of a particular project, minimizing risks and contributing to effective application development

Andrei Chmelev

This article reviews contemporary architectural approaches to building real-time web applications that require instant data updates. It highlights a comparative analysis of key technologies—WebSockets, Server-Sent Events (SSE), and WebRTC—providing guidance on selecting the most suitable solution for scenarios such as chat systems, online editors, and multiplayer games. In addition, it explores strategies for scaling and load balancing real-time connections, as well as mechanisms to ensure reliable data delivery. Typical architectures—such as Node.js with Redis Pub/Sub or Kafka—are presented, along with recommended methods for monitoring and logging real-time data streams. Code snippets illustrate practical implementation techniques to help developers create robust and efficient real-time features within various infrastructures.

Bulycheva Mariia 

The article discusses the development of graph neural network (GNN) models for e-commerce, aimed at predicting user interactions with the content of the main page. The methods described focus on utilizing networks to improve relevance and personalize the user experience. The purpose of this work is to examine the features of the graph neural network architecture specifically designed for e-commerce tasks. This architecture operates on graph data structures, allowing for the consideration of different levels of connections between users and products, and their various features. The methodology employs graph node embedding algorithms such as GraphSAGE and Node2Vec, which transform any data into numerical vector representations. The sources used include scientific articles published by the author in the public domain, as well as materials available on the Internet, enabling a comprehensive exploration of the topic. The results demonstrate that the proposed architecture enhances the accuracy of content personalization, as evidenced by increased click-through rates (CTR) and revenue. The implementation of this system modifies the algorithms for ranking content, thereby impacting the platform’s effectiveness. The article will be valuable for specialists working on recommendation systems, researchers, and developers of e-commerce solutions. The conclusions affirm the success of the proposed architecture in data analysis and user experience adaptation.

Di Feng

The article examines modern financial systems, which are a multi-layered structure where data processing requires the use of analytical methods. The relevance of the topic is due to the need to develop algorithms that take into account the complexity of information and the volume of data. The main focus of the article is on the use of machine learning algorithms for managing credit risks and assessing their adaptation to financial activities. The article discusses approaches aimed at predicting risks in credit institutions. The emphasis is on models adapted to the specifics of financial processes, taking into account atypical situations. This allows you to create tools that meet the requirements of the industry. The use of algorithms increases the accuracy of forecasting borrowers’ creditworthiness, detects fraud, and improves loan portfolio management processes. Methods related to the identification of relationships between data characteristics contribute to the development of more effective models. Such approaches demonstrate resilience to changing financial market conditions. The conclusions of the article are addressed to specialists involved in risk management, developers of analytical solutions, and researchers in the field of economic applications of machine learning. The use of these tools helps to optimize the internal processes of credit institutions and increase their efficiency in working with data. The use of machine learning algorithms opens up prospects for accurate analysis of credit risks. These technologies improve the quality of forecasting, strengthen the sustainability of organizations, and create conditions for successful adaptation to changes in the financial environment.

Sandeep Singh 

The article discusses approaches to improving identity and access management, adapted to modern challenges related to technology development. The research is aimed at identifying the problems of traditional methods and developing solutions based on modern tools, including artificial intelligence, cloud services, and the concept of complete abandonment of trust. The article also analyzes the transformation of the architecture of identity and access management systems (hereinafter - IAM) in the context of technological changes. IAM components adapt to the challenges posed by the introduction of new solutions and technologies. Technological development, including cloud platforms, artificial intelligence algorithms, and innovative systems, has changed the architecture of IAM. It began to adapt to distributed environments, cyber threats, and new attack scenarios. In turn, the creation of platforms with flexibility, security, and adaptability helps organizations protect information resources, confront new threats, and maintain sustainability in the digital world. The results of the study confirm the ability of the proposed methods to ensure data protection, minimize the likelihood of attacks, and comply with regulatory requirements. The materials are intended for cybersecurity specialists, access control system developers, and employees working with digital transformation. The conclusion focuses on the effectiveness of the proposed approaches to the organization of access control that meet modern requirements of the digital environment, strengthen information protection, and facilitate adaptation to new challenges.

Farzon Nosiri 

This study evaluates the performance of five next-generation sequencing (NGS) read mapping tools—BWA, Bowtie, Stampy, Mosaik, and Agile—focusing on their behavior under varying conditions, including read errors, insertions, deletions, and mutation rates. Reads from Illumina sequencing systems were simulated using dwgsim and further manipulated for controlled testing. The evaluation assessed mapping precision, sensitivity, and computational efficiency using a 2.7 GB reference genome to ensure compatibility across tools. Stampy demonstrated superior overall performance, particularly with complex variations, but exhibited limitations in handling perfect reads with base quality “!” scores. Bowtie and Mosaik performed well under specific conditions but struggled with paired reads and precision, respectively. Agile proved incompatible with short reads, while BWA emerged as the fastest tool, albeit with precision issues at higher mutation rates. Deletions posed the most significant challenge across all tools. These findings highlight critical trade-offs between speed, accuracy, and adaptability in NGS read mapping tools, providing insights for their application in population genomics and other domains.