Utilized LLMs and generative AI for interactive 3D content generation. My missions were:- State-of-the-Art Research: Conducted a thorough review of the latest research and technological advancements in LLMs and VLMs to guide the project direction.- Model Testing and Fine-Tuning: Fine-tuned and evaluated LLMs and VLMs for better understanding of 3D worlds using advanced techniques such as Low-Rank Adaptation (LoRA) and Representation-Efficient Fine-Tuning (REFT).- Implementation: Developed and implemented multiple models using PyTorch, the Hugging Face Transformers library, and Streamlit.3D Space Creation: Applied multimodal AI to generate and interpret complex 3D living environments, enhancing the functionality of virtual spaces.- Technology Monitoring: Kept up to date with relevant research and important technologies throughout the internship by regularly analyzing new publications and innovations in the field.

- Exploration of super-resolution methods to improve the quality of thermal images. Development of a CNN model to detect cracks in multispectral images.- Use of PyTorch, OpenCv, Sklearn.- Publication under review: Z.Lambert, R.Antoine, C.Mauget, "Image-based crack detection on badlands using visible and thermal infrared classification".

- Development of a calibration tool for an epidemiological model using real data.- Integration of the Pandas, Numpy and LMFIT libraries.- Integration into the EpiPolicy project, to help calibrate prediction models.

15-hour training by Google on the basics of machine learning. Practical application using Tensorflow APIs

- Developed a freelance business as a 3D artist.- Use of Blender, Technical and Commercial Communication, Creativity- More than 15 international clients and 30 projects.