Looks like you have JavaScript disabled. For the full Twine experience, you will need to re-enable it.

Hi, I’m Frederick Spreafichi, a Machine Learning Engineer with a MSc in Mathematics and hands-on experience in ML/DL using PyTorch, GPU training on HPC (Slurm/CUDA), and web/API development. I enjoy turning research into scalable, production-ready systems and collaborating with cross-disciplinary teams. My experience includes an ML research internship at Area Science Park, building graph-augmented neural cellular automata, and developing reproducible training pipelines on GPU clusters. I’m also experienced in front-end web development, delivering responsive UI features and robust APIs in collaboration with designers and back-end engineers.…Hi, I’m Frederick Spreafichi, a Machine Learning Engineer with a MSc in Mathematics and hands-on experience in ML/DL using PyTorch, GPU training on HPC (Slurm/CUDA), and web/API development. I enjoy turning research into scalable, production-ready systems and collaborating with cross-disciplinary teams. My experience includes an ML research internship at Area Science Park, building graph-augmented neural cellular automata, and developing reproducible training pipelines on GPU clusters. I’m also experienced in front-end web development, delivering responsive UI features and robust APIs in collaboration with designers and back-end engineers.

Frederick Spreafichi

AI Engineer, Data Scientist, Web Developer, +2





Hi, I’m Frederick Spreafichi, a Machine Learning Engineer with a MSc in Mathematics and hands-on experience in ML/DL using PyTorch, GPU training on HPC (Slurm/CUDA), and web/API development. I enjoy turning research into scalable, production-ready systems and collaborating with cross-disciplinary teams. My experience includes an ML research internship at Area Science Park, building graph-augmented neural cellular automata, and developing reproducible training pipelines on GPU clusters. I’m also experienced in front-end web development, delivering responsive UI features and robust APIs in collaboration with designers and back-end engineers.…Hi, I’m Frederick Spreafichi, a Machine Learning Engineer with a MSc in Mathematics and hands-on experience in ML/DL using PyTorch, GPU training on HPC (Slurm/CUDA), and web/API development. I enjoy turning research into scalable, production-ready systems and collaborating with cross-disciplinary teams. My experience includes an ML research internship at Area Science Park, building graph-augmented neural cellular automata, and developing reproducible training pipelines on GPU clusters. I’m also experienced in front-end web development, delivering responsive UI features and robust APIs in collaboration with designers and back-end engineers.

Available to hire

Hi, I’m Frederick Spreafichi, a Machine Learning Engineer with a MSc in Mathematics and hands-on experience in ML/DL using PyTorch, GPU training on HPC (Slurm/CUDA), and web/API development. I enjoy turning research into scalable, production-ready systems and collaborating with cross-disciplinary teams.

My experience includes an ML research internship at Area Science Park, building graph-augmented neural cellular automata, and developing reproducible training pipelines on GPU clusters. I’m also experienced in front-end web development, delivering responsive UI features and robust APIs in collaboration with designers and back-end engineers.

Skills

Experience Level

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Intermediate

Language

Italian

Fluent

English

Fluent

German

Intermediate

Work Experience

Deep Learning Research Intern at Area Science Park

June 1, 2025 - September 1, 2025

Built a graph-augmented Neural Cellular Automata in PyTorch on a 40 ×40 canvas with 16 channels and sparse mid-range attention (radius 4, 8 sampled neighbors, message every 3 steps); lightweight model with 11,185 parameters(+27% vs classic NCA). Engineered a reproducible training pipeline (pool training 1024 states, batch 16, 800 steps/epoch; rollouts 48–80 steps with 40% long rollouts 200–400), with checkpointing every 5 epochs and TensorBoard logging for ablation-style iteration. Trained for 960 epochs on GPU (Slurm) and achieved best-case reconstruction quality of SSIM 0.934, PSNR 32.53 dB, and Pixel Perfection 0.838 on the target pattern.

Front-end Web Developer at Ideo Network

October 1, 2019 - December 1, 2020

Delivered responsive UI features with HTML/CSS/JavaScript, improving cross-browser behavior and page performance in a delivery-oriented workflow with designers and back-end developers.