Home

Howdy! My name is Shantanu Thorat, and I am a University of Cambridge Master's grad. I completed my BSc in Computer Science (with Engineering and University honors) at Texas A&M University with a minor in economics. I was a receipient of the merit-based President's Endowed Scholarship at A&M for all four years. In my spare time, I enjoy reading, playing basketball and chess, and visting natural and historic landmarks. My favorite places that I have visited so far include Hawaii, Switzerland, and London.

• Email: st980 [at] cantab.ac.uk
• LinkedIn
• GitHub
• HuggingFace
• Google Scholar
• ORCID

News

2025

• September: Released source code for the DACTYL project
• August
  • Released preprint for my Cambridge master's thesis DACTYL: Diverse Adversarial Corpus of Texts Yielded from Large Language Models
  • Published dactyl-generation on pypi, a Python package that streamlines batch inference across multiple LLM API providers
• June: Won 1st place in the Cambridge-McKinsey Risk Prize essay competition for my essay, "Did ChatGPT Write This? An Exploration of the Risks of AI-Generated Texts in the Wild". Read the essay here.

2024

• December: Presented Which LLMs are Difficult to Detect? A Detailed Analysis of Potential Factors Contributing to Difficulties in LLM Text Detection as poster at NeurIPS 2024 Safe Generative AI Workshop

• September: Mössbauer and EPR detection of iron trafficking kinetics and possibly labile iron pools in whole Saccharomyces cerevisiae cells published in Journal of Biological Chemistry

2023

• December
  • A kinetic model of iron trafficking in growing Saccharomyces cerevisiae cells; applying mathematical methods to minimize the problem of sparse data and generate viable autoregulatory mechanisms published in PLOS Computational Biology
  • 1st place team at Texas A&M's CSE Capstone Fall 2023 Expo