I’m an AI and cybersecurity engineer, researcher, and computer science educator. I enjoy connecting ideas and data to solve practical problems—and making the reasoning clear along the way.
Bachelor of Computer Science (Honours), with distinction University of Windsor · Artificial Intelligence specialization · Minor in Mathematics
I’m drawn to problems where the pieces exist, but the right system hasn’t been put together yet. I enjoy understanding an environment, finding useful connections in its data, and turning an idea into something people can use.
I’m an independent consultant and computer science educator, with experience in AI research, security analytics, industrial cybersecurity, and technical instruction. I’m also interested in graduate research in AI and cybersecurity, building toward a Master’s and PhD.
My consulting work also includes manual LLM red teaming and Python/API workflows for website classification. I enjoy making the process inspectable through real samples, clear documentation, and practical evaluation.
An international exchange at Université de Tours gave me the chance to work on AGORA and strengthen my French. I work in English, French, and Arabic, and enjoy collaborating across cultures.
Work with me
↗AI & data systems Retrieval, agent workflows, model integration, and data preparation.
↗Cybersecurity Network analytics, security labs, technical writing, and training.
↗Teaching & collaboration One-to-one CS tutoring, project guidance, and research opportunities.
Some things I’ve worked on
Page from the original workResearch prototype · Jan–Apr 2026
AGORA — Agent-Based Government Open-Data Reasoning Assistant
Designed and implemented AGORA, the orchestration component of the co-authored Agent-Enhanced Data Services (AEDS) framework, to answer natural-language questions over public datasets through retrieval, planning, and specialized execution paths.
Designed hybrid retrieval over dataset metadata with vector search and BM25.
Separated general retrieval-based reasoning from technical, computation-oriented processing.
Structured the system for explainable, attributed, data-driven answers.
Added evaluation workflows for discovery and analytical questions across multiple model configurations.
The local-PC demo query took approximately 2 minutes 30 seconds. The prototype downloads selected resources and processes them sequentially. Parallel agents, caching, and selective API access are proposed improvements; their speedup has not been measured. Evaluation remains preliminary.
Actual application · Browse interface screenshotsPersonal Project
Ressist — AI-Assisted Resume & Cover Letter Platform
Full-stack application that stores a structured career profile in a vector database and generates
tailored cover letters and resumes for job posts, grounded in the user’s own
experience rather than generic templates.
Built a FastAPI backend with hybrid (vector + BM25) retrieval over Weaviate and OpenAI embeddings.
Used DSPy for structured extraction from free-form experience text and RLM-style resume generation with tool-based retrieval.
Shipped a React + Vite dashboard for CRUD, career profiles, testimonials, generation history, and JSON backup export.
A client pipeline connecting network telemetry to graph structures for behavioral analysis. The delivered work covers data ingestion and graph construction; GNN-based detection was a future application.
Wrote a practical guide and Enterprise ATT&CK reference material to help security teams connect tactics, techniques, and procedures with defensive workflows.
Explore more work Research, consulting projects & coursework
Original screen recording · Context-based adversarial testingConsulting engagement · May 2025
LLM Red Teaming — Adversarial Prompt Testing
Designed and ran manual adversarial prompts against ChatGPT, creating 13 recorded examples with chat logs. Tested cybersecurity scenarios and authorization framing, including a case where a refusal changed to security-related guidance after a claim of red-team authorization.
LLM red teamingAdversarial promptingSafety evaluationTechnical documentation
Documented the prompt sequence, refusal behaviour, and subsequent model responses through screen recordings and transcripts.
Built a Python workflow using the Claude Sonnet API to classify websites as Indie, Corporate, or Unsure and produce structured CSV output. Iterated on prompt engineering, batch processing, webpage-content retrieval, token limits, and model selection to balance processing time, API cost, and classification quality.
The workflow was developed for a large domain list, with a brief targeting up to one million websites. The work shown here is sample output from the development runs.
Original team paper · Model training curvesAcademic team research · Feb–May 2025
Lung Tissue Image Classification
Co-authored a team study of CNNs for classifying histopathology images into benign tissue, adenocarcinoma, and squamous cell carcinoma. Academic image-classification research; not a clinically validated diagnostic system.
Explored convolutional neural networks and systematic hyper-parameter search to improve detection performance on imaging data.
Combined ensemble methods and data augmentation to strengthen robustness across training conditions.
Documented methodology and results suitable for research-style reporting and technical review.
Deep LearningTensorFlowPythonJupyter NotebookResearch Proposals
Prepared more than 1.5 TB of data for a treatment-cost prediction project, using Python, Pandas, and Azure to handle fragmented inputs and processing beyond local-PC capacity.
Page from the original workSecurity research · University of Windsor
Reproducible Network Security Lab
Built and documented a physical and virtual environment for authorized attack simulation and traffic collection, combining Windows/Linux VMs, VLANs, routing, and automated packet capture.
Completed a privacy-focused messaging prototype with Python, MetaMask, and a web interface. Our team won the $500 HyperCycle.AI Challenge (Option A): Build Privacy Preserving Social Reputation.
Explored LSTM-based classification of Arabic tweets into positive, negative, neutral, and objective categories. Reviewed Arabic NLP research and documented preprocessing and language-specific challenges.
Arabic NLPLSTMTensorFlow/Keras
Advanced networking coursework · Feb–May 2025
Secure Data Transfer Protocol
Implemented a custom data-transfer protocol while applying networking and security concepts in coursework.
NetworkingProtocol design
Hackathon team project · Hack the Valley
KnowYourCulture
Contributed a Flask and MongoDB backend to an application for learning about cultures and traditions.
FlaskMongoDBBackend development
Canadian Nuclear Laboratories · May–Aug 2024
Industrial Cybersecurity Training
Built simulated industrial systems and practical exercises connecting attack demonstrations to monitoring, investigation, and defensive action.
Designed and implemented AGORA during a research exchange at Université de Tours. I was responsible for the orchestration component of the co-authored AEDS framework, building a working prototype for questions over France’s open-government data catalogue.
Designed planning, retrieval, dataset routing, execution, and answer synthesis around approximately 73,000 catalogue records.
Built streamed progress and token/cost tracking, and studied discovery and analytical questions across model configurations.
AI & Cybersecurity Consultant
Independent · Upwork & direct engagements
Sep 2024 – Present
I help clients turn technical requirements into usable systems, clear documentation, and reliable data workflows.
AI content generation: designed a private client application using fine-tuned OpenAI models and controlled orchestration to meet writing-style and page-format requirements (Mar–Jul 2025).
LLM red teaming: designed adversarial prompts and documented 13 examples through screen recordings and chat logs, testing cybersecurity scenarios and authorization framing (May 2025).
Website classification: built a Python/Claude API workflow with structured CSV output, prompt iteration, batch processing, and API cost optimization for a large domain list (Oct 2025).
Healthcare data preparation: processed more than 1.5 TB of scattered data using Python, Pandas, and Azure for a treatment-cost prediction project (Dec 2024–Mar 2025). My contribution was data preparation.
Cybersecurity writing: prepared a practitioner-focused MITRE ATT&CK guide and reference material connecting security concepts to SOC workflows (Mar 2025).
Privileged access management: researched and prepared an educational presentation on PAM and cloud access security for a client (Sep 2024).
Cybersecurity / ML Systems Consultant
Athena Security Group (formerly Athena Software Group) • Remote
May 2025 – Jun 2025
Designed a pipeline that converts mirrored network traffic into graph data for security analysis.
Connected SPAN traffic capture, Dockerized Zeek, Filebeat, Logstash, and Neo4j to ingest connection, DNS, and HTTP logs.
Wrote Ruby ETL transformations that generated Cypher updates, implementing connection-as-node and connection-as-edge graph models for behavioral network analysis.
Documented deployment, troubleshooting, and remaining engineering work. Graph-based anomaly detection was a future application.
Independent Computer Science Educator
Self-employed · Windsor & remote
May 2025 – Present
I adapt each lesson to the student’s goals and current understanding, using questions, visual explanations, code tracing, and guided debugging.
Teach programming, data structures, algorithms, systems programming, operating systems, software engineering, and related CS subjects through individual lessons and project mentoring.
Full-stack capstone mentoring · May–Aug 2025: guided a Wayne State University student on an inherited election-data web application developed for General Motors. This was a tutoring engagement with the student.
Supported GitHub onboarding, Docker setup, React pages, Flask API integration, database administration through pgAdmin, and U.S. government open-data collection.
Provided guidance for data collection and planning a future linear regression model to examine electoral trends.
Cyber Security Researcher / Intern
Canadian Nuclear Laboratories • New Brunswick, Canada
May 2024 – Aug 2024
Worked in cybersecurity training and simulation environments with strong emphasis on
blue-team instruction, incident response, and industrial OT/IT security.
Built VMware ESXi/GNS3 industrial-control simulations with PLCs, HMIs, and engineering workstations for reproducible attack and incident-response exercises.
Integrated Tenable, Nozomi, and Elastic Stack into SOC training workflows for industrial OT monitoring and investigation.
Configured FortiGate firewall rules and Cisco traffic mirroring to support monitoring and defensive controls.
Developed and delivered hands-on cybersecurity training for nuclear energy workers, connecting attack demonstrations to threat detection, investigation, and mitigation.
Created MITRE ATT&CK-aligned attack modules, blue-team playbooks, attack cards, and step-by-step exercises, including reconnaissance, keystroke injection, and Bash-based response workflows.
Researched configuration-management approaches, including SaltStack, for managing devices in the simulated environment.
Cyber Consulting Intern
CEC · SEPCO3 / General Electric project · Cairo, Egypt
Jun 2023 – Aug 2023
Supported cybersecurity planning from Cairo for the Al-Wahda power plant project in Morocco. CEC was a subcontractor to SEPCO3, which worked under General Electric (GE), the company leading the power plant project.
Communicated regularly with SEPCO3’s Chinese project team and analyzed extensive power-plant documentation provided through the project.
Analyzed RTU security requirements for the Al-Wahda power plant project in Morocco.
Compared vendor RTU security features and procurement specifications; documented communication risks, firewall/VPN restrictions, and proposed network protections.
Produced network diagrams and technical recommendations for management.
Research Assistant — Network Security
WASPLabs, University of Windsor • Windsor, Canada
Oct 2022 – Apr 2023
Built distributed security lab environments for attack simulation, packet capture, and dataset-oriented experimentation.
Built a penetration-testing lab with 3 physical hosts, 6 Windows/Linux VMs, and 3 VLANs, documenting installation, routing, and connectivity tests in an 11-page setup guide.
Simulated MITM, DoS, cache-poisoning, and session-hijacking attacks using Python/Scapy and C packet tooling; analyzed traffic with Wireshark, tshark, and tcpdump.
Automated traffic capture at startup with Bash for cybersecurity dataset collection and machine-learning research.
Created Docker-based cloud lab environments on Google Compute Engine and AWS.
IT Intern
Asfour Crystal · IT Department · Jul–Aug 2022
Gained hands-on exposure to manufacturing IT across networking, hardware support, enterprise software, and data-center operations, including VLAN segmentation and Oracle ERP.
A sample of my teaching for high school and early university computer science students. Step-by-step examples and whiteboard annotations explain the “why” and connect the details to the bigger picture.
Fall 2023 · COMP 2540 — Data Structures & Algorithms
Office hours, coursework assessment, feedback, and explanations that connect code to underlying ideas.
University tutoring · Appointed Oct 2023
CS General Tutor
Helped University of Windsor students develop confidence in programming, algorithms, systems, and related computer science topics. Across tutoring and TA work, I have supported more than 50 students.
Independent teaching & mentoring · May 2025–Present
Understand the reasoning, not just the answer
One-to-one support across CS year levels: C, Java, Python, Bash, data structures and algorithms, Unix/Linux systems programming, operating systems, digital logic, computer architecture, and software engineering.
From an inherited codebase to independent problem-solving
Guided a Wayne State student on a capstone developed for General Motors: an election-data web application using React, Flask, Docker, and government open data. Supported implementation and data collection, and planning for a future regression model.
People I’ve worked with
Client Feedback
Emily Conrad — Children’s book author
“Salman was great to work with - he is both personable and professional. He made himself available (despite time differences) and was able to help me with my project. Highly recommended!”
CollaborativeProfessional
Executive Feedback
Peter Worth — CEO, Athena Security Group
“Solid resource, self motivated and skilled in Cyber Security analysis and development”
Committed to Quality
Capstone Mentorship
Thomas K. — Wayne State University
“Working with Salman as my tutor for my final computer science project was a great experience. He guided me through the full stack development process, step by step, and took the time to make sure I understood each concept thoroughly.”
Read the full reference
“Working with Salman as my tutor for my final computer science project was a great experience. He guided me through the full stack development process, step by step, and took the time to make sure I understood each concept thoroughly. He was always willing to meet whenever he could, and it was clear he genuinely cared about helping me succeed. Salman went above and beyond, not just explaining what to do, but why it was done that way which made a huge difference in my learning. Overall, he's a great tutor and an even better person. I really appreciate all the help and support he gave me throughout the project.”
Anna Shi — Cyber Security Research Scientist, Canadian Nuclear Laboratories
“During his internship, Salman displayed great enthusiasm and adaptability in his work. He took on challenging, open-ended tasks that required him to learn new concepts and then quickly apply them.”
Read the full reference
"Salman was a student at Canadian Nuclear Laboratories from May 2024 to August 2024. During his internship, Salman displayed great enthusiasm and adaptability in his work. He took on challenging, open-ended tasks that required him to learn new concepts and then quickly apply them. For example, Salman needed to quickly learn computer networking and then immediately apply the concepts to the setting up of a virtual network on a hypervisor. He was also flexible and accommodating, such as when he pivoted to working on a demonstrative cyber attack that was needed for a training course. On the personal side, Salman was always friendly and quickly integrated into the team. He became friends with the other students and formed a tight bond with them."
Personal reference
Eman El Shazly — Lead Manufacturing Controls Engineer, Stellantis
Personal character reference supplied for my CNL application.
Salman possesses excellent communication and interpersonal skills. His ability to listen actively and communicate effectively contributes to a harmonious work environment.
Read the full reference text
I hope this letter finds you well. I am writing to provide a character reference for Salman Youssef, who I have had the pleasure of knowing for the past 10 years. I believe that Salman would be an excellent fit for the opportunity at your company due to his outstanding character and professional qualities.
Over the past decade, I have witnessed Salman consistently demonstrate qualities that makes him an asset in both personal and professional settings. One of Salman's most notable attributes is his integrity. In all our interactions, I have seen him adhere to a strong moral, making decisions based on honesty and fairness.
Salman is also a dedicated and hardworking individual. Whether it's in his personal or professional pursuits, he approaches tasks with diligence and commitment. This work ethic, coupled with his ability to stay organized and focused, allows Salman to consistently deliver high-quality results.
Salman possesses excellent communication and interpersonal skills. His ability to listen actively and communicate effectively contributes to a harmonious work environment.
Furthermore, Salman is known for his resilience. I have seen him navigate through challenging situations with grace and maintain a positive attitude even under pressure. This ability to stay calm and collected in the face of adversity makes Salman an invaluable team member.
In conclusion, based on my 10-year association with Salman, I am confident that his outstanding character, work ethic, and interpersonal skills makes him an excellent candidate for the opportunity at your esteemed company. I believe Salman would not only meet but exceed your expectations.
Read more student reviews Original TutorOcean screenshots and complete quotations
Student reviews are reproduced verbatim in their original English. The screenshots preserve the source wording and ratings.
TutorOcean review
Chukwudi O.
Excellent tutor for Computer Science
Salman is a fantastic tutor. He explains difficult concepts in a clear, simple way and is very patient when you need things repeated. His lessons are well organized, and he uses plenty of examples that make topics click. He's friendly, professional, and always on time. My confidence and understanding improved a lot after working with him. Highly recommended!
TutorOcean review
Senior A.
Amazing tutor
I met Salman at the university during my first year and ever since that day he helped me pass all my computer science courses I had difficulties with he always made sure I understand everything so I can succeed and he still tutors me right now as I’m entering my final year
TutorOcean review
Youssef D.
The gold standard for Computer Science tutoring, highly recommended!
Salman possesses an incredible depth of knowledge in Computer Science, but what makes him truly exceptional is his ability to listen. He listens carefully to where you're struggling, meets you right at your skill level, and breaks down complex technical concepts into clear, intuitive building blocks. Patient, deeply skilled, and genuinely invested in your growth. I cannot recommend him enough!
TutorOcean review
Omar K.
Knowledgeable, Dedicated, and Supportive friend and tutor
I had the pleasure of studying alongside Salman throughout our years at university, taking several computer science courses together, including COMP 1410, COMP 2120, COMP 2560, COMP 2650, COMP 2660, COMP 3110, and others. Throughout our studies, Salman consistently demonstrated a strong commitment to learning and keeping up with course material. What stood out most was his willingness to help others, often organizing group study sessions before midterms to review course content, explain difficult concepts, and help us prepare for exams.
Based on my experience studying with Salman, I can confidently say that he is knowledgeable, dedicated, and great at helping others understand challenging topics. His patience, willingness to share his knowledge, and ability to work through course material with others make him an excellent tutor. I would highly recommend Salman to any student looking for help with computer science.
ETL design, Neo4j graph construction, Zeek telemetry workflows,
Logstash processing, and real-time pipeline thinking for analytics and detection.
Delivery Mindset
Work structured around clarity, maintainability, practical constraints,
architecture decisions, and polished output suitable for clients and stakeholders.
Technical Instruction
Experience tutoring computer science, supporting university coursework,
and designing hands-on cybersecurity training that turns complex topics
into structured, practical learning.
Research & Evaluation
Experimental design, model and pipeline evaluation, technical reporting,
reproducible prototyping, and evidence-driven iteration across AI and ML projects.
