# Research — Josué Godeme

- Computer Science, Dartmouth College
- King Scholar & Stamps Scholar
- Incoming Schwarzman Scholar, Tsinghua University (2027)

I research and build at the intersection of multilingual AI, voice technologies, and education. My work spans evaluating LLM performance across languages, deploying models on edge devices for low-connectivity classrooms, and designing voice-first learning systems. I also ship products — most recently [Orphi](/projects/orphi), an AI language companion live on the App Store.

Growing up between Adja, French, and English — and now learning Mandarin — shaped how I think about language technology. Most AI systems are built for a handful of languages and fail everyone else. I am interested in changing that: not just through linguistics research, but by building the infrastructure and evaluation frameworks that make multilingual AI actually work at scale.

## Areas

- Multilingual NLP and LLM evaluation across languages.
- Edge-deployed AI for low-connectivity educational contexts.
- Voice AI and speech technologies.
- AI infrastructure for underrepresented languages.

## Publications

- Adja-French Parallel Corpus: A New Resource for Machine Translation of a West African Under-Resourced Language — Godeme, J. et al. (2025). LREC 2026.
- Artificial Allies: Validation of Synthetic Text for Peer Support Tools through Data Augmentation in NLP Model Development — Godeme, J., Hill, J., Gaughan, S. P., Hirschbuhl, W. J., Emerson, A. J., Darabos, C., Bobak, C. A., & Fortuna, K. L. (2025). Proceedings of the Pacific Symposium on Biocomputing, Hawaii, USA.

## Talks

- Pacific Symposium on Biocomputing, Hawaii, USA (Jan 2025)

## Training

- Bluedot Impact, Technical AI Safety — Completed technical track on AI alignment, governance and evaluation — applied to multilingual and edge AI.