Denzel Chiu

Kigali, Rwanda

Denzel Chiu

Denzel Chiuseni

Cybersecurity | Software Engineering | Adversarial AI

Currently researching polymorphic malware at CMU-Africa

I build secure software and study how AI systems fail under attack. Graduate Research Associate at Carnegie Mellon University Africa, with an MSc in Information Technology specializing in cybersecurity.

View projectsView CV

About

I work where cybersecurity, software engineering and adversarial AI overlap. My training is in security, starting with a BTech in Information Security and Assurance at Harare Institute of Technology and continuing with an MSc in Information Technology at Carnegie Mellon University Africa in Kigali.

I also build production software. At Huza HR I engineered Node.js and TypeScript APIs and microservices, and my own projects include PhishGuard, a phishing simulation platform, and Wireshark Agent Suite, which uses an LLM to classify network traffic and raise alerts.

My research asks how AI systems hold up when someone is actively trying to break them. My paper at ISBM 2026 compares how classical and transformer based smishing detectors fail under adversarial attack, and a follow up study looks at the same problem in English and Swahili, where training data is scarce. As a Graduate Research Associate at CMU-Africa I also research malware behaviour and run self hosted LLM infrastructure.

I'm open to roles in security engineering, software engineering or AI security, and I do my best work on teams where those areas meet.

Cybersecurity

Vulnerability assessment and penetration testing, IT controls review, phishing simulation and network traffic analysis.

Explore

Software Engineering

Full stack systems in Python and TypeScript: REST APIs, microservices and Next.js interfaces, deployed on Azure, Heroku and Vercel.

Explore

Adversarial AI

Adversarial robustness of ML detectors, with published research on smishing detection and self hosted LLM deployment with Ollama and vLLM.

Explore

Experience

  1. February 2026 to Present

    Graduate Research Associate

    Carnegie Mellon University Africa, Kigali

    • Build and operate a multi backend LLM environment serving quantized Qwen models with Ollama and vLLM across local hardware, Azure and rented GPU servers.
    • Research malware behaviour and threat detection under Prof. Jema David Ndibwile, and write up findings and recommendations for faculty and project stakeholders.
    • Collaborate with faculty on cybersecurity projects across the African technology landscape, documenting methods, results and next steps.

  2. August 2025 to December 2025

    Graduate Teaching Assistant, Ethical Hacking

    Carnegie Mellon University Africa

    • Supported students through penetration testing, exploitation and lab exercises.

  3. June 2025 to December 2025

    Full Stack Developer

    Huza HR, Kigali

    • Engineered scalable RESTful APIs with Node.js and TypeScript, using PostgreSQL for data management and retrieval.
    • Built and maintained a TypeScript microservices architecture that supports HR automation workflows.
    • Designed responsive Next.js interfaces for recruiters and job seekers.

  4. January 2024 to July 2024

    IT Support Staff

    Avante Screens Pvt Ltd, Harare

    • Deployed digital signage across 15 retail stores in 8 locations.
    • Implemented web based content management systems to speed up data processing.

  5. December 2021 to July 2022

    Software Developer Intern

    CICOSY Pvt Ltd, Harare

    • Developed and tested web applications with Spring Boot and Angular.
    • Took part in code reviews and fixed bugs to keep systems reliable.

Education

MSc in Information Technology (Cybersecurity)

Carnegie Mellon University, Pittsburgh, Pennsylvania

August 2024 to December 2025 · GPA 3.88 / 4.0

BTech in Information Security and Assurance

Harare Institute of Technology, Harare

August 2019 to May 2023 · First Class with Distinction, top 10%

Selected work

Security · AI

Wireshark Agent Suite

AI powered network traffic analysis tool that captures live traffic or reads pcap files with tshark, classifies flows with the Gemini API and raises severity based alerts with reasoning.

  • Python
  • FastAPI
  • WebSockets
  • SQLite
  • tshark
  • Gemini API
  • Next.js
  • TypeScript
GitHub

Security · Web · AI

PhishGuard (Baitline)

Phishing awareness training platform that runs sandboxed phishing simulations for organisations and individual users.

  • Next.js
  • TypeScript
  • React Query
  • Zod
  • JWT
  • Express
Live demo

Security · Web

Password Auditor

Flask API deployed on Heroku that checks passwords and emails against HIBP Pwned Passwords and XposedOrNot using k-anonymity, and returns a binary access flag with breach exposure details.

  • Python
  • Flask
  • Heroku
GitHub

Security

UPMS Vulnerability Assessment and Penetration Test

Assessed IT general controls across the Upanzi Programs Management System, found critical weaknesses and recommended fixes. January 2025 to May 2025.

  • Burp Suite
  • Nmap
  • Metasploit

Security · AI

Agentic Bug Bounty Hunting System

Architecture for an autonomous bug bounty assistant.

  • Ollama
  • Qwen
  • Gemini

AI

Multi Backend LLM Environment

Quantized Qwen deployment across local, Azure and Vast.ai hardware.

  • Ollama
  • vLLM
  • Azure
  • Vast.ai

Publications

  1. [1]

    “Adversarial Robustness in Smishing Detection: A Comparative Analysis of Adversarial Fragility in Classical vs. Transformer Based Detection Systems”

    Denzel Chiuseni, Athanase Bahizire, Silva Hama, Jema David Ndibwile

    5th World Conference on Information Systems for Business Management (ISBM 2026), Bangkok, September 2026. Springer Nature (LNNS).

  2. [2]

    “Architectural Fragility in Low Resource Smishing Detection”

    English and Swahili smishing detection robustness study.

    In preparation for PeerJ.

Skills

Programming and Software Engineering
PythonTypeScriptJavaScriptNode.jsJava (Spring Boot)AngularNext.jsREST APIsmicroservicesGitGitHubDockerJest
AI and Data
adversarial machine learningadversarial robustness evaluationLarge language modelsagentic AIprompt engineeringmodel quantizationself hosted inference (Ollama, vLLM)Claude CodeCursorKiro
Cloud and Platforms
AzureHerokuVercelLinux (Bash)PostgreSQLMySQLSQLite
Security
Vulnerability assessment and penetration testingIT general controls reviewaccess and identity managementincident responsethreat detectionBurp SuiteMetasploitNmapWiresharkSnortSuricata

Contact

Open to roles in security engineering, software engineering and AI security, as well as freelance work and research collaboration.