Here is a brief overview of my work and interests. I am interested in building AI systems that can reason but that are also safe and fair.
Research
I was lucky enough to see my work published in top-tier conferences (ICML, ICLR, AAAI, ACL, NAACL, Nature Medicine). Some of my work has been featured in press (Nature, Nature Machine Intelligence, Washington Post).
I am an Editor for the Neuro-symbolic AI journal and an Action Editor for the ACL Rolling Review.
Highlights:
- 📚 TextGrad: Automatic “Differentiation” via Text
- 📚 How Well Can LLMs Negotiate?
- 📚 Lessons from Safety-tuned LLaMAs
Open Projects
My projects have collected thousands of stars on GitHub: Contextualized Topic Models, FashionCLIP, TextGrad, Reclist, PLIP. You can check my GitHub profile if you want to know more.
Talks
I gave several talks to companies and universities. Some of them are available on YouTube: Pinecone, Outerbounds, LightOn AI, Neptune AI.
Blog Posts
Other than papers, I have also written several blog posts on various topics: