Ricardo Fitas
I am currently working as a Research Assistant. I specialize in applying data science, artificial intelligence, and numerical optimization techniques to real-world problems. With a strong foundation in Mechanical Engineering, Supply Chain Management and Data Science, I am well-prepared to tackle complex problems, drive innovation, and contribute meaningfully to scientific and technological progress.
Throughout my career, I have been dedicated to helping companies and individuals grow by leveraging my expertise. I have had the privilege of working as a research associate at the Technical University of Munich, where I gained valuable experience in various aspects of engineering and data science. I am also certified by the Massachusetts Institute of Technology (MIT). At TU Darmstadt, I have significantly increased research capacity, even in situations of low resources available.
I have supervised more than 8 BSc and MSc student theses, and reviewed for Elsevier and Springer Nature more than 100 times. I am also part of the Technical Program Committee of IEEE conferences.
Projects I’m Interested In
Neuro-symbolic AI for Materials Science
Combining domain constraints with learned representations to accelerate discovery and improve interpretability in materials design tasks.
Optimization-driven Design of Corrugated Structures
End-to-end pipeline for topology/parameter optimization of corrugated components under multi-objective constraints (weight, stiffness, manufacturability).
ML for Manufacturing Quality
Predictive models and SPC-style monitoring for defect detection and process drift, with a focus on low-data regimes.
Training Optimizers with Distillation & Transformers
One challenge in my MO-ETPSO method is the lack of dynamic dimensionality adaptation during iterations. I’m exploring transformer-based policy distillation to learn when/how to vary the search-space dimensionality over time. Ongoing work.
Agentic AI for Open Scientific Problems
Exploring agent frameworks that help non-experts collaborate with AI to make progress on open research questions (decomposition, tooling, verification).
Site-wide RAG for My Work
Improve my website assistant so it can answer from my code, projects, and papers without asking me—using retrieval over GitHub, PDFs, and notes.
AI-Driven Package Recommendation (Pareto Front)
Build a recommendation system that ranks packaging solutions on a learned Pareto front (weight and stiffness). Users in industry can pick the “best for them” option based on market and supply-chain context; supports trade-off exploration and what-if scenarios.
Agentic AI for Automated Web Design
Use agentic AI to plan, prototype, and iterate website UX/UI automatically—covering layout generation, color/typography selection, accessibility checks, and performance budgets. The goal is a loop where agents propose designs, test them (Lighthouse/axe), and refine based on metrics and user goals.
Future Projects
Doctoral Thesis Focus
Plan to shift more time toward formal doctoral research and coursework.
YouTube Channel for Research
Create a channel to share talks, seminars, and explainers of my papers to a wider audience.