Siyona Sarma
Computer Science, UC Berkeley
Email: siyona [DOT] sarma [AT] berkeley [DOT] edu
People
Mentors:
I am actively seeking Summer 2026 internships!
Feel free to reach out via email or LinkedIn.
About Me
I am an undergraduate student at UC Berkeley majoring in computer science. My research interests include Large Language models' text analysis abilites,
building data systems powered by LLMs, and the development of comprehensive global datasets to improve the performance of various machine learning models. I also
am working on the development of a new department, Human Technology Futures, as a part of the UCB College of Computing, Data Science and Society.
Outside of the classroom, I am the President of the Political Computer Science club, which addresses social issues like food scarcity, reproductive health,
and more through computational and data science driven methods.
In my free time, I am the social media manager at the German Shepherd Rescue of Northern California and a volunteer at Guide Dogs for the Blind.
Research
EPIC Data Lab
Creating a splitting algorithm for 1000+ page unstructured texts, using tools like OCR and GPT;
improving DocETL performance with this technology, academic paper in progress
Responsible AI Initiative at BAIR (Berkeley AI Research) Lab
Led a workshop of 20+ academics ideating about data cooperative and data governance using speculative design methods, academic paper in-progress
Building a global image dataset focused on gender.
(past) Experimental and Computational Linguistics Ensemble (ECOLE) Lab at SF State
Developed different prompt engineering techniques for text simplification using various LLM models including Chat GPT 3.5 and 4.0
Gained a mastery on syntactic complexity and the features of text complexity when analyzing ML model text simplification
Utilized CTAP (Common Text Analysis Platform) to visualize how text is simplified by grade level in Newsela corpora
Presented research project at the Data Discovery symposium at UC Berkeley and to other local colleges
Publications
Text Simplification for Children: Evaluating LLMs vis-à-vis Human Experts
ACM CHI Conference 2025, Stanford Undergraduate Research Conference 2025
Email: siyona [DOT] sarma [AT] berkeley [DOT] edu
LinkedIn, GitHub, Google Scholar, etc.