Software Engineering, AI & Security

Mohamed Aziz Makhlouf

M.Sc. Computer Science — AI Specialization. Building secure AI systems, scalable backends, and practical full-stack applications. Entwicklung sicherer KI-Systeme, skalierbarer Backends und praxisnaher Full-Stack-Anwendungen. Développement de systèmes d'IA sécurisés, de backends évolutifs et d'applications full-stack concrètes.

My engineering philosophy is simple: turn intelligent ideas into reliable software.

I combine a strong computer-science foundation with hands-on experience in artificial intelligence, backend development, web applications, data analysis, and security-oriented evaluation. At TU Darmstadt, I have worked on RAG-based fact-checking, agentic AI security and full-stack development. I enjoy translating complex technical problems into maintainable systems that can be tested, explained, and improved.

Technical Arsenal

AI & Machine Learning

LLMs, NLP & Intelligent Systems

PythonPyTorchscikit-learnpandasNumPyNLTKSentence TransformersLLMsRAGPrompt EngineeringDeepfake DetectionAgentic AI Security

Backend & Data

APIs, Services & Databases

JavaSpring BootFastAPIExpress.jsREST APIsPostgreSQLMongoDBSQLSocket.IOAuthenticationAPI Integration

Frontend & Applications

Responsive Interfaces & Full-Stack Apps

TypeScriptJavaScriptAngularReactHTMLCSSResponsive UI

DevOps & Engineering

Delivery, Testing & Collaboration

DockerDocker ComposeGitGitHub WorkflowsPostmanLinuxTestingCode ReviewsAgile Development

Featured Engineering Work

AI-Powered Cyber Incident Platform

Full-Stack Security Engineering

Designed and implemented a platform for submitting, managing, and tracking cybersecurity incidents. The system combines secure authentication, structured incident workflows, persistent storage, and AI-assisted analyst support in a full-stack architecture.

View Architecture & Code
AngularTypeScriptSpring BootJavaPostgreSQLREST APIsDockerAI APIs

APR. 2026 — Present

Systematic Analysis of Agentic AI Systems

Agentic AI Security Research

Reproducing and analyzing agentic AI systems to evaluate how reliably LLM agents follow goals, use tools, and respond to adversarial inputs. Implementing direct and indirect prompt-injection attacks, reproducing defense mechanisms, and comparing their effectiveness in controlled experiments.

Python Langchain LLM AgentsDirect Prompt InjectionIndirect Prompt InjectionvLLMAdversarial EvaluationLinux

Apr. 2026 — Present

Data Analysis of Scratch Programming Communities

Data Analysis and Machine Learning

Built a Python pipeline to collect, clean, structure, and analyze Scratch forum threads and replies. Combined topic clustering, problem-outcome labeling, engagement metrics, and emotion and support analysis to study how children discuss programming difficulties and help each other solve them.

PythonpandasBeautifulSoupNLTKscikit-learnSentence TransformersLLMsMatplotlib

Jun. 2025 — Dec. 2025

Automated Multimodal Fact-Checking

Bachelor Thesis · Grade 1.3

Extended an automated multimodal fact-checking system with deepfake-detection and metadata-analysis modules. Developed an LLM-based metadata reasoning component using ExifTool and OpenAI API endpoints, integrated the resulting evidence into the RAG pipeline, and evaluated the system on benchmark datasets in a Linux compute environment.

Results: increased DEFAME's accuracy from 50% to 63% across nearly 1,000 DGM4 claim–image pairs, while improving recall for refuted claims from 37% to 59%. On the custom Metaset, metadata verification raised accuracy from 74% to 78% and refuted-claim recall from 66% to 82%.

Measured Impact on Fact-Checking Performance

PythonPyTorchExifToolOpenAI APIsRAGLLMsPrompt EngineeringLinux

Nov. 2024 — Feb. 2025

Multimodal Fact-Checking Web Platform

Agile Full-Stack Development · Grade 1.0

Built a full-stack web interface for multimodal fact-checking with text and image inputs. Integrated live fact-checking responses through and contributed to testing, code reviews, collaborative GitHub workflows and Docker-based deployment.

View Architecture & Code
ReactJavaScriptExpress.jsFastAPIPythonMongoDBSocket.IODockerGit

Teaching Experience

As a tutor at TU Darmstadt, I supported students across core computer-science subjects by explaining complex concepts clearly, guiding structured problem solving, and providing practical feedback.

2024 — 2025

Software Engineering

Tutor · TU Darmstadt

Supported students with software-development methods, software design, and collaborative programming, helping them translate engineering principles into structured project work.

Software Design Development Methods Team Programming

2023 — 2024

Algorithms & Data Structures

Tutor · TU Darmstadt

Taught algorithms, data structures, complexity analysis, and systematic problem solving, using Java to connect theoretical concepts with practical implementations.

Algorithms Data Structures Complexity Analysis Java

2023 — 2024

Propositional & Predicate Logic

Tutor · TU Darmstadt

Guided students through formal logic, proof techniques, and logical reasoning, helping them develop precise and well-structured approaches to abstract problems.

Formal Logic Proof Techniques Logical Reasoning

Education

2025 — Present

M.Sc. Computer Science — AI Specialization

Technical University of Darmstadt

Grade: 1.26

Focus: Artificial Intelligence, Deep Learning, NLP, probabilistic models

2022 — 2025

B.Sc. Computer Science

Technical University of Darmstadt

Grade: 1.54

Focus: Software engineering, algorithms, AI, databases, and web development

Honors & Awards

National Academic Distinction

12th Place Nationwide

Tunisia · 2021

Achieved the 12th-highest national ranking in the Tunisian Baccalaureate, Mathematics specialization.

Mathematics National Ranking Academic Excellence