I’ve worked on all kinds of projects, but at heart I’m a creative researcher and problem solver who uses technology and data to study how technology shapes society, and how new tools can be used to make things better.
These days, I’m focused on AI, especially AI sycophancy, activation steering, uncertainty in reasoning and information retrieval. (Special thanks to Claude for advancing this work)
Before that, I worked at USAID as a Senior Data & Analytics Advisor, where I focused on global food security and the Feed the Future initiative. Prior to USAID, I was at the Fire Department in New York City (FDNY) as Director of Operations Research which overlaped with the COVID-19 pandemic giving me an incredible opportunity to quantitatively support the Deaprtment's emergency response.
Earned my PhD in 2018 where I focused on applying ML to understand and optimize cities, including waste forecasting, real-time urban census, and ML for low-cost air quality monitoring.
- From Thought to Answer
- Max Fink Retrieval Augmented Generation
- The Geometry of Persona
- Prompt Fragility in TruthfulQA
I'm using different AI tools to learn about specific NLP and AI concepts. One might argue this is a really bad idea, but I do like saying "hmm... I dont understand this..now build me some python code so I can learn by doing..."