Resume

Universität des Saarlandes 2025 - current

Ph.D. student in Computer Science

Advisor - Dr. Michael Hahn

Universität des Saarlandes 2023 - 2025

M.Sc. in Computational Linguistics

GPA - 1.1 (4.0 on US scale)

Manipal Institute of Technology 2016 - 2020

B.Tech Computer Science and Engineering

CGPA - 9.09 / 10

Research Assistant in Language, Computation and Cognition Lab Dec 2023 - Sep 2025

Theoretical analysis of the strengths and limitations of all neural network building blocks
Studying Length Generalisation in Transformers

ML Engineer, Glib.ai Dec 2020 - Aug 2023

Lead developer on one of the flagship products Finray, a financial statement analyser
Implemented novel models for automated extraction of relevant tables from long documents

Intern, Samsung R&D Bangalore Jan 2020 - Jun 2020

Intern, Novartis India Ltd Nov 2018 - Jul 2019

Veitsman, Y., Jobanputra, M., Sarrof, Y., Bakalova, A., Demberg, V., Pavlick, E., & Hahn, M. (2025). Born a Transformer–Always a Transformer?. NeurIPS 2025
Huang, X., Yang, A., Bhattamishra, S., Sarrof, Y., Krebs, A., Zhou, H., Nakkiran, P., & Hahn, M. (2025). A Formal Framework for Understanding Length Generalization in Transformers. ICLR 2025
Sarrof, Y (2025). Homophonic Pun Generation in Code Mixed Hindi English, 1st Computational Humour Workshop, COLING 2025
Sarrof, Y., Veitsman, Y., & Hahn, M. (2024). The Expressive Capacity of State Space Models: A Formal Language Perspective. NeurIPS, 2024

LCT Summer School 2025 Prompting Workshop
Computational Humour Workshop, COLING 2025 Homophonic Pun Generation in Code Mixed Hindi English
Formal Languages and Neural Networks seminar (FLaNN), 2024 Expressive Capacity of SSMs
TaCoS, Student Conference, 2024 Transformers or RNNs or SSMs: Who’s more sensitive?

Reviewer : NeurIPS 2025
Organizer :
- LangLunches, a biweekly research talk series at Saarland University
- NeurIPS ELLIS Poster Session, Saarbrücken, 2025
- NeurIPS ELLIS Poster Session, Saarbrücken, 2024
Student Council Member, Department of Language Science and Technology, Saarland University

Top Reviewer, NeurIPS 2025: Recognized among the top 10% of reviewers of the conference.
Masters with Distinction: Awarded for achieving an overall grade of 1.1; distinction is reserved for cumulative grades of 1.1 or lower (1.0 is the highest possible grade).
Winner, Smart India Hackathon 2019: Led a 6-member team to 1st place in a national 36-hour hackathon involving 300,000 participants. Developed an AI-driven game suite to assess children’s cognitive skills.
Finalist, GE HackElt Pan India 2019: Selected as a top 10 national finalist for developing a predictive hospital management system to reduce patient wait times using data analytics.

Tags