"Our technology forces us to live mythically."
— Marshall McLuhan
Hi there! I'm Luca. I asked ChatGPT how to describe myself and it said I'm a 'Creative Technologist.' I thought it fit. Really, I do all sorts of things; I'm a writer, filmmaker, ML researcher, and, when time allows, an entrepreneur. I'm pursuing a master's in Computer Science at Stanford with a concentration in AI, while developing films that explore how modern technologies shape identity, relationships, and well-being.
I'm currently writing a feature for 20th Century Studios (Disney) and recently directed Silk Road, a first look based on my original script. In the lab, I work on world-modeling and physically grounded perception inspired by how humans learn—methods that reduce reliance on labels and move toward richer causal understanding.
We live in an age where, each day, technology redefines what we thought possible. In this sea of possibilities, I strive to chart new paths. I build systems that push technology toward empowering outcomes, and I write stories that help us reckon with—and reimagine—their impact on our lives.
Stanford University • M.S. Computer ScienceGraduate Researcher — Stanford Neuroscience & Artificial Intelligence Laboratory
GPA: 4.12
Stanford University • B.S. Computer ScienceCURIS Research Fellowship
CS 109 class competition winner — Naïve Bayes credit re-attribution for AI-written dialogue
Psychological horror feature screenplay co-written with Thomas Wheeler. Currently in development at 20th Century Studios (Disney).
Introduced SpelkeNet, a model for discovering physically grounded object segments that move together under causal forces.
First solo feature screenplay and companion 'First Look' short film, both currently in development.
Research project from the Stanford Neuroscience & Artificial Intelligence Lab (Yamins Lab). Focused on developing a physically controllable world model (PhyWM) that supports zero-shot object understanding. Currently under review at ICLR 2026.
Created a probabilistic model to attribute AI-generated TV dialogue back to original shows with 99.7% accuracy.
MAE/ViT model that fuses satellite data to generate counterfactual terrains for open-world games.
Social venture promoting consent education via a condom brand marketed to colleges, blending product design and advocacy.
Built an interactive game that uses a CNN to classify ASL gestures in real-time, reaching 96% accuracy across 70k+ images.
Refined Swin V2 Transformers for 4× super-resolution of human faces, improving reconstruction of details and pigmentation.
Startup streaming service for fantasy sports, featuring original viewing value algorithms and WME representation.
A psychological horror spec about neuroscience, possession, and identity.
|
I'm always interested in connecting with fellow creatives, researchers, and builders. Whether you're working on a film project, AI research, or product development, feel free to reach out.
No promises on speed of response — I get so many emails from Walgreens I may have to go off the grid.