Université de SherbrookeAcademic project
SemEval 2025: classifying food hazards and extracting associated passages
The multi-task model reaches 0.7937 on the ST1 metric and 0.7606 on the metric combining classification with passage extraction.
Engineering student in Artificial Intelligence and Data Science
Adrien Skrzypczak
I work on AI and data to automate repetitive tasks, and to turn a real need into something I can answer with data.
Paris + Hauts-de-France
Projects
Internships, coursework, and personal experiments: the context, the choices made, the results, and what remains to be improved.
Université de Sherbrooke
Nicolaudie GroupProfessional project
RAG prototype for structured configurations
Turning a request into structured configuration while retrieving the right technical documentation.
- Université de Sherbrooke
Academic project
NASA NEO: comparison of six classifiers on an imbalanced dataset
SVM: 99.43% recall, but 29.9% precision on this imbalanced dataset.
Personal project
PairWise Cards: spaced repetition flashcard application
Deployed local-first application: three review modes, SM-2 scheduling, and persistent progress without an account.
Education & experience
My background in computing and AI/data, and my first professional experiences.
Education
Master's degree in AI & Data Science
Université de Sherbrooke
2025-2026 · Sherbrooke, Quebec
Courses in machine learning and deep learning, natural language processing, probabilistic models, recommendation, and statistics.
Education
Computer engineering degree, data/software track
EILCO
2023-2026 · Hauts-de-France
Computer engineering curriculum covering software development, machine learning, signal and image processing, and technical project management.
Experience
Data/AI Analyst Intern — Research team
AI Builders
September 2026 - February 2027 · Paris
Research on LLMs and agentic AI, comparison of data/AI solutions, proof-of-concept development, and internal knowledge sharing. The internship focuses on studying what is technically mature, useful for a use case, and presentable to both consultants and their clients.
Experience
Generative AI internship - RAG prototype
Nicolaudie Group
June-August 2025 · Montpellier
Design and evaluation of a prototype capable of transforming a natural-language request into structured configuration for professional software. Work on technical documentation, the RAG pipeline, output validation, the API, and containerization.
About
An engineering student in a dual EILCO and Université de Sherbrooke program, I keep this portfolio as a project notebook: what I tried, what I learned, what worked, and what I want to improve. I came to AI and data by trying to automate tasks and answer real questions with data.
How I work
Before coding, I try to understand the problem and the expected outcome. I test several approaches, compare their results, and iterate based on what I observe.
What I want to build
I like experimenting, but I don't want to stop at the prototype. What matters to me is that an AI or data project can actually be used, not just demoed: that means working on the data, the code, the evaluation, and keeping things simple enough to evolve. That's also why topics like MLOps or AIOps interest me, even though they're not part of my current internship, and my interests will probably keep evolving with experience.
Notes
Reflections on my experiments, what I am learning, and the things I try outside my main projects.
Contact
Currently interning at AI Builders in Paris until the end of February 2027. I remain open to discussions about AI, data, and my projects.