cv

Table of contents

Basics

Name Benjamin V. Rozonoyer
Label PhD Candidate in Machine Learning and NLP
Email brozonoyer@umass.edu
Url https://brozonoyer.github.io
Summary PhD candidate at UMass Amherst (IESL) working on discrete diffusion models for text, with broader interests in representation learning, efficient LLM inference, and generative information retrieval.

Work

  • 2025.06 - 2025.09
    Research Assistant
    Oracle Labs
    MoE LLM inference optimization, Machine Learning Research Group.
  • 2025.02 - 2025.06
    Student Researcher
    Google Research
    Generative ranking for information retrieval (New York).
  • 2024.05 - 2024.08
    Research Assistant
    Oracle Labs
    LLM agents for automated program repair.
  • 2023.06 - 2023.08
    Scientist Intern
    Charles River Analytics
    Research internship in Cambridge, MA.
    • Transformer-based contrastive learning for authorship embeddings (IARPA HIATUS) — advanced CRA from last to 2nd place in evaluation
    • Probabilistic programming in Pyro for generation/inference of military activity patterns
  • 2022.09 - Present
    PhD Student
    University of Massachusetts, Amherst
    Doctoral research in machine learning and NLP, advised by Andrew McCallum.
    • Discrete diffusion models for text — non-autoregressive, flexible, and controllable language generation (thesis focus)
    • Generative ranking for information retrieval
    • Efficient inference with mixture-of-expert LLMs
    • Geometric representation learning for large directed graphs
  • 2020.06 - 2022.08
    Engineer II
    Raytheon BBN Technologies
    Software engineer and researcher in information extraction and NLP.
    • Multilingual information extraction (DARPA AIDA)
    • Full NLP stack for COVID-19 machine reading system (IARPA COVID Seedling)
    • Maintained internal NLP and deep learning libraries; integrated open-source and research models
  • 2019.06 - 2019.08
    Research Intern (TechStudent)
    Raytheon BBN Technologies
    Multilingual neural relation extraction via machine translation and MGIZA++ token-level annotation projection to train mBERT.
  • 2017.08 - 2019.12
    Teaching Assistant
    Brandeis University
    TA for five undergraduate and graduate courses.
    • Statistical Approaches to NLP (Fall 2019)
    • Phonological Theory (Spring 2019)
    • Data Structures and Algorithms (Fall 2018)
    • Theory of Computation (Spring 2018)
    • Discrete Structures (Fall 2017)
  • 2017.01 - 2021.09
    Research Assistant
    Brandeis University — BiRCh Project
    Morphological analysis, parsing, segmentation, tokenization, transcription, and adjudication for the Bilingual Russian Children corpus.

Education

  • 2022 - Present
    PhD
    University of Massachusetts, Amherst
    Computer Science
    • Machine Learning
    • Distributed and Operating Systems
    • Advanced Algorithms
    • Probabilistic Graphical Models
    • Reinforcement Learning
  • 2020 - 2021
    MS
    Brandeis University
    Computational Linguistics
  • 2016 - 2020
    BS
    Brandeis University
    Computer Science, Linguistics, and Mathematics

Awards

Certificates

Skills

Machine Learning
PyTorch
JAX
Transformers
Diffusion Models
Representation Learning
LLMs
CUDA
Natural Language Processing
Information Extraction
Information Retrieval
Computational Linguistics
Multilingual NLP
Programming
Python
Julia
Java
SPARQL
CUDA C++

Languages

English
Native
Russian
Native