Alvin Sarkadi

This paper focuses on investigating Mandiant APT 1, a well-known Advanced Persistent Threat (APT) linked to Chinese state-sponsored cyber-espionage, and explores how machine learning and data analytics can be applied to prevent such sophisticated attacks. APTs represent a persistent threat to organizations by infiltrating networks over long periods, exfiltrating data, and causing severe security breaches. By analyzing the tactics, techniques, and procedures (TTPs) used by APT 1, the paper evaluates the effectiveness of advanced technologies like machine learning models, anomaly detection, and big data analytics in identifying patterns, predicting attacks, and mitigating risks in real-time. This research underscores the importance of automating threat detection and response mechanisms to counter the evolving landscape of APTs and highlights the role of machine learning in building proactive defenses against these sophisticated cyber adversaries.

This research project delves into the evolving landscape of cybersecurity threats, with a particular focus on the high-profile APT1 incident attributed to Mandiant. Our group has explored how this sophisticated cyber attack could have been mitigated at various points in time through the strategic use of available network resources.

1. Intent-Based Networking (IBN) Description: IBN allows administrators to define high-level goals (intents) for the network, which are then automatically translated into policies. The system continuously monitors and adjusts the network to meet these goals, using AI for automation and proactive management. IBN simplifies network operations by aligning them with business objectives. 2. Software-Defined Networking (SDN) Description: SDN separates the control plane from the data plane, providing centralized, programmable control of network resources. It allows for dynamic, automated management of network traffic, improving flexibility and scalability, and is widely used in areas like cloud computing and network virtualization. Key Difference: IBN is goal-oriented and uses AI to achieve business-driven outcomes, while SDN focuses on centralizing and automating network control through architectural separation.

The two projects focus on virtualization within the AWS platform, specifically involving the creation and configuration of multiple EC2 instances. These projects include the setup of PuTTY for secure access, along with a comprehensive explanation of IP addressing and subnetting.

This research focuses on Endpoint Security within a Microsoft 365 (M365) environment, encompassing the management, configuration, and updating of policies, logs, and related aspects. The study explores strategies and best practices for securing endpoints effectively, leveraging M365 tools and features to establish robust security frameworks. Key considerations include policy formulation, configuration management, log monitoring, and ongoing updates to mitigate risks and enhance security posture in modern organizational settings.

This document delves into the configuration and implementation of diverse security policies within the M365 E5 environment. It offers an extensive guide detailing the deployment of specific security features tailored to M365 tenants.

Zero Trust is a security concept predicated on the principle of continuous verification and strict access controls, regardless of the location of users or resources within a network. When applied to Microsoft 365 (M365), Zero Trust can significantly bolster the security posture of the tenant by systematically verifying identities, scrutinizing device health, and monitoring access requests in real-time. This approach minimizes the risk of unauthorized access and potential breaches, enhancing overall resilience against evolving cybersecurity threats within the M365 ecosystem.

The presentation delivered a clear and professional message tailored effectively to the Information Technology leadership team. It included essential slides with appropriate technical depth, complemented by insightful use of presentation notes, a practice increasingly observed in IT presentations from major corporations like Microsoft. The thorough coverage of each slide, especially highlighting 'Compliance Policies,' provided a detailed overview of crucial security measures such as conditional access and risk policies within the M365 environment, emphasizing their pivotal role in fortifying organizational security against potential threats.

The presentation explores the comprehensive features of Microsoft Intune and elucidates how Tetra Shillings Accounting benefits from leveraging Intune for configuration and management purposes.

Achieved an overall score of 87%, with 87 correct answers out of 100. Here's a breakdown of my performance across different lessons: Managing Authentication: 13/14 correct Installing and Updating Windows 10: 19/24 correct Managing Devices: 16/18 correct Managing Security: 11/13 correct Configuring Auditing: 14/14 correct Planning and Managing Microsoft Intune: 14/17 correct My exam duration was 1 hour and 20 minutes, and the exam was taken on June 24, 2024.

This project conducted a comparative analysis between IBM Watson and Snap Logic for Natural Language Processing (NLP) applications, focusing on sentiment analysis of textual data. Using two distinct input texts—one positive and one negative restaurant review—the study explored how each platform analyzes sentiment and emotion within text. IBM Watson's NLP capabilities provided detailed sentiment scores for individual keywords and overall text emotion, showcasing its robust analytical depth. In contrast, Snap Logic offered simpler polarity scores suitable for larger datasets, emphasizing practicality in scenarios like analyzing extensive restaurant reviews. The project also utilized FreeTTS for text-to-speech conversion and Google Translate for round-trip translations of technical excerpts, highlighting their effectiveness despite occasional grammatical nuances. Overall, the study underscores the potential of NLP AI tools in enhancing various industries, from hospitality to healthcare, by facilitating better customer service, multilingual support, and streamlined communication processes.

This project explores the application of Artificial Intelligence (AI) and Machine Learning (ML) in sentiment analysis using datasets sourced from John Hopkins University's Multi-domain Sentimental Analysis repository. The chosen dataset includes pre-processed tarballs featuring reviews from diverse domains such as books, DVDs, electronics, and kitchen products. Utilizing tools like IBM Watson and Snap Logic, the project analyzes sentiment and emotion within these datasets to enhance understanding of user experiences and opinions across various electronic products. Additionally, the project involves the implementation of Amazon Lex for creating a chatbot designed to facilitate car rental interactions. Using prompt slots for user inputs such as location, rental duration, name, and phone number, the chatbot guides users through the rental process and validates inputs based on predefined utterances and slot types. This demonstrates Amazon Lex's capability to handle structured dialogues and potentially automate reservation processes. The project underscores the practical applications of AI/ML in interpreting consumer sentiment for product improvement and customer service enhancements in retail and service industries.

This Computer Vision project explores the capabilities of Amazon Rekognition, an AI service provided by AWS, through various image analysis scenarios. The project, conducted as part of CMSC 307 - Artificial Intelligence Applications, focuses on evaluating Rekognition's effectiveness in object and celebrity recognition, as well as its potential applications in public safety and quality control. The project begins with an overview of navigating and using Amazon Rekognition via the AWS Management Console. Several features are highlighted, including label detection, which analyzes images to identify objects, people, and environments. The project systematically evaluates Rekognition's performance across different image types: Urban Environment Analysis: A New York City street image is used to assess label detection's accuracy. While successful in identifying several objects and environments with high confidence scores (e.g., city, intersection), there were limitations such as mislabeling objects and people, which suggests room for improvement in complex urban scenes. Wildlife Image Analysis: An image featuring multiple animals in a savanna is analyzed to test Rekognition's ability to correctly identify animals and their ecosystem. The results showed accurate identifications of zebra, elephant, and antelope/impala, though with varying confidence scores. However, some expected labels like specific tree types were not identified, indicating potential categorization challenges. Indoor Kitchen Scene Analysis: An indoor kitchen image is uploaded to test Rekognition's ability to recognize indoor environments and kitchen appliances. While it successfully identified the setting as a kitchen and recognized a few objects like plants and electrical devices, the confidence scores for many object labels were unexpectedly low, highlighting challenges in object categorization. Celebrity Recognition: Rekognition's celebrity recognition feature is tested using images of public figures like Lewis Hamilton and Jensen Huang. It successfully identified Jensen Huang with high confidence but struggled with a lookalike image of Lewis Hamilton, demonstrating the sensitivity of recognition accuracy to image quality and context. The project concludes by discussing potential applications of Amazon Rekognition in industries such as public safety and quality control. It suggests Rekognition's utility in law enforcement for crime scene analysis and in industries like pharmaceuticals for quality inspection, despite identified limitations in complex image categorization and labeling. Overall, the project showcases the capabilities and limitations of Amazon Rekognition in various image analysis tasks, emphasizing its potential impact across diverse industries through enhanced object and celebrity recognition, albeit with ongoing challenges in accuracy and contextual understanding.

Further explains the implementation of cloud computing at Don & Associates.

Don & Associates have taken close consideration to moving their current IT infrastructure to a cloud computing infrastructure due to cost and efficiency.

10 Capture the Flag Challenges

Handling the Physical security, and network security of a hospital, maternity ward, entrances and exits. Furthermore, maintaining the network infrastructure in an efficient matter to track unusual behavior on the hospitals Wi-Fi network.

Describes the current network vulnerabilities at the local hospital.

This paper compares and analyzes the storyline of Roald Doal's 'Charlie and the Chocolate Factory' to Mel Stuart's version of the film and Tim Burton's version of the film

This paper discusses the use of FRT (Facial Recognition Technology) and how it has played a critical role in China, and the war in protests in Hong Kong.

Connecting 4 workstations to 1 switch, then to 1 router. The cabling and commands used to create this network are in the PDF. This project was used with Cisco Packet Tracer to create the scenario.