Taha Draoui

AI & Software Engineer | PhD Student

Taha Draoui portrait

About Me

Results-driven AI and software engineer with hands-on experience building LLM-powered systems, agentic workflows, and machine learning pipelines. Strong interest in developing intelligent, autonomous systems that bridge research and real-world usability, with a focus on code quality, automation, and reliable ML engineering.

Python LangChain LangGraph PyTorch Scikit-Learn NumPy Pandas SQL LLMs Docker Git CI/CD

Experience

Graduate Research Assistant (AI / ML)

University of Michigan

Sep 2024 – Present
  • Developed an LLM-driven system to detect and correct preprocessing and overlap data leakage in Python ML code.
  • Leveraged a custom program slicing framework to isolate each model’s training and evaluation pipeline, reducing unnecessary context for LLM analysis.
  • Improved detection F1 score by 21% for preprocessing leakage and 15% for overlap leakage over prior state-of-the-art.
  • Worked at the intersection of static code analysis, machine learning engineering, and agentic AI systems.

Software Engineer Intern – Backend & Systems

SWIFTECH

May 2024 – Aug 2024
  • Contributed to backend API development as part of a cross-functional engineering team.
  • Implemented CI/CD pipelines to automate testing and deployment.
  • Worked across backend systems, ASP.NET Core, React.js, and agile development workflows.

Education

PhD in Computing

University of Michigan–Flint

Expected 2027
  • Fully funded PhD student (GSRA – full tuition coverage)
  • Global Merit Scholarship — awarded for academic excellence

Master of Science in Computer Science

University of Michigan–Flint

2024 – 2025
  • 3.88 GPA
  • Graduate Scholarship (Non-Resident Tuition Award)
  • Graduate Student Research Assistant (GSRA)
  • Graduate Student Instructor (GSI)

Publications

Analyzing Vulnerability Overestimation in the Maven Ecosystem

MSR 2025 – Mining Challenge Track

May 2025
  • Taha Draoui, Faten Jebari, Chawki Ben Slimen, Munjaap Uppal, Mohamed Wiem Mkaouer

Programs and Fellowships

NSF I-Corps (Regional), Participant

National Science Foundation

2025
  • Completed customer discovery and market validation using the NSF I-Corps methodology.

INSPIRE Student Scholar (Inaugural Cohort)

University of Michigan–Flint

2025
  • Selected through competitive nomination recognizing academic excellence and leadership potential.