AZIZ BEN ALI
I’m AZIZ BEN ALI, a graduate research and teaching assistant focused on building intelligent systems that can learn and generalize across real-world environments.
About
I’m AZIZ BEN ALI, a graduate research and teaching assistant focused on building intelligent systems that can learn and generalize across real-world environments. My work spans cognitive agents, cross-environment transfer learning, and generative AI for scientific data—ranging from collecting and fine-tuning trajectories in computer-use and Minecraft to training diffusion and equivariant models for 3D atomic structure generation.
Beyond research, I enjoy translating complex ideas into usable outcomes: developing custom training and sampling pipelines, validating model outputs with domain experts, and supporting learning through Python tutoring tailored for engineering data analysis. I also bring industry experience from roles in deep learning and data science, where I built transformer-based trajectory prediction, developed anomaly detection support, and prototyped RAG-based assistants—always with an emphasis on measurable performance, reliability, and clarity.
- 3 years' experience
- Joined October 2026
Skills· 19
Languages
- EnglishFluent
- GermanFluent
- FrenchAdvanced
- ArabicFluent
Industry experience
- Computers & Electronics
- Education
- Energy & Utilities
- Professional Services
- Software & Internet
- Manufacturing
- Telecommunications
Work experience· 10 roles
Worked on MCP-Agent: a unified framework for building cognitive agents across heterogeneous environments. Reviewed agent frameworks for computer-use and Minecraft, collected agent trajectories, and fine-tuned multiple Qwen/Qwen3 model sizes with LoRA. Also performed joint fine-tuning to evaluate cross-environment transfer learning.
Developed an E(3)-equivariant graph neural network for rotationally and translationally invariant 3D atomic structure generation. Designed and trained diffusion-based generative models (DDPM/DDIM) for atomic coordinates and implemented custom sampling pipelines producing CIF files. Collaborated with computational chemistry researchers to validate generated structures. Supported tutoring for Python for engineering data analysis—connecting machine learning concepts with visualization and practical workflows.
Developed an E(3)-equivariant graph neural network for rotationally and translationally invariant 3D atomic structure generation and implemented diffusion models (DDPM/DDIM) for atomic coordinates. Built custom sampling pipelines to produce CIF files and collaborated with computational chemistry researchers for validation. Tutored Python for engineering data analysis with ML-to-visualization workflows.
Education
Hire AZIZ BEN ALI today
Post a job and invite AZIZ, or contact them directly. No hidden commission.