Yada Pruksachatkun

I do research on multimodal reasoning and model behavior steering. Some of my previous notable work has been in transfer learning with SuperGLUE and work in responsible AI.

I am driven by the goal of making models think the way humans do. AI is increasingly being developed and used as copilots to people, as companions, as therapists, and as assistants to experts.

Let's raise the bar for what interacting with AI feels like.

Timeline:

2018-2021/2: NLP Research

  • After learning about machine learning during internships in college, I decided to try to understand these models more in depth. At the time, Natural Language Processing was really starting to taking off with this new architecture called Transformers. My first research project was at Microsoft Research, where I investigated using ML to detect moments of change in mental health forums with Amit Sharma.
  • I then spent several wonderful years at NYU, where I was fortunate to be advised by Sam Bowman on transfer learning, including work with SuperGLUE and transfer-learning toolkit jiant.
  • After my masters, I worked at Amazon Alexa's responsible AI team (now Amazon Responsible AGI) as a scientist.

2022-2025: Startups

  • Itching to build for end customers, and having always wanted to build early-stage startups, I took the plunge and learned how to build 0-to-1 for high stakes usecases, automating phone calls between payors and providers at Infinitus and building agents in recruiting at Moonhub. These years were invaluable in learning what it takes to take new technology to market and change human workflows.
  • Of course, I wasn't able to fully stay disconnected from research for long, so also contributed to BLOOM and a couple other research projects throughout the years. I also had the opportunity to start and run the PhD research internship program at Infinitus, where we conducted work on dialogue systems that were accepted at top-tier NLP conferences.

2025: Return to Research

The ML world looks a lot different than in 2022. After acquisition of my last startup to Salesforce, I'm currently at Salesforce Research's Interactive AI Team, researching multimodal reasoning and model behavior steering/emotional intelligence.

Looking forward, I am excited to do good research and work on capabilities alignment in models until we reach AGI.

I have only been able to do the things I have done due to people taking a bet on me. If you are an underrepresented minority, or want to collaborate, feel free to book time here.