ABOUT
Hi, I’m Zoe.
A technical founder building emotionally intelligent AI.
I’m interested in how AI models represent emotion and express it through language, voice, and multimodal interaction. I explore these questions by building systems, reading research, and designing small experiments.
At the workbench
I’ve built a voice agent, a content agent for creating videos, and fine-tuned a small text-to-speech model. I’m also exploring new models and training a wake-word model.
What you’ll find here
Research is where I develop focused questions, connect the literature, and outline experiments others can reproduce. TIL is a broader notebook of practical lessons from building a company as a technical founder. Projects brings the systems themselves together.
Current questions
- What should an emotion representation capture?
- How can a small speech model express emotion with useful control?
- How should language, voice, and visual context work together?
- What would convincing evaluation look like?
Find me
You can find my public code and connect with me on GitHub ↗.