I am a Postdoctoral A.I. Research Scientist at Lawrence Livermore National Laboratory. Current research includes video processing for computer vision and multimodal LLMs. Research and mentoring for video processing and computational objectness from video. Amodal segmentation and content completion from challenging, cluttered video scenes.

Developing a library for adapting arbitrary vision and language models to be multimodal for improved transfer to special domains. This research uses specialized lightweight adapter modules alongside low-rank parameter-efficient finetuning (LoRA, QDoRA) Future work will adapt fine-tuned domain expert LLMs and domain-specific data encoders to bring multimodality to the physical sciences. I am writing occasional updates about the design and implmentation of this research here.

Previously advised by Professor Randy Paffenroth for a PhD in Machine Learning 2019-2023.

My research spans computer vision for video processing, LLMs / foundation models, stable fine-tuning, and transferrable representation learning for deep neural networks to improve real-world performance.

Select Publications and Projects

Google Scholar

• An Analysis of Segment Anything 2
  • Probing weaknesses of state-of-the-art video object segmentation model to find potential improvements for obscured object tracking.
• Addressing Issues with Working Memory in Video Object Segmentation
  • Modifying the working memory buffer in VOS models (SAM2, XMem, Cutie) can benefit the tracking of objects during intermission periods between relevant information.
• ChemVise: Maximizing Out-of-Distribution Chemical Detection with the Novel Application of Zero-Shot Learning
  • International Conference on Machine Learning and Applications 2023
  • Oral presentation and paper
• ACGANs Improve Chemical Sensors for Challenging Distributions
  • International Conference on Machine Learning and Applications 2022
  • Oral presentation and paper
• Cycles Improve Conditional Generators
  • International Conference on Advanced Data Mining and Applications 2022
  • Paper and poster
• Identifying Struggling Students by Comparing Online Tutor Clickstreams
  • Artificial Intelligence in Education 2021
  • Short paper and poster
• Identifying Explanations Within Student-Tutor Chatlogs
  • International Conference on Educational Data Mining 2022
• ChemTime: Rapid and Early Classification for Multivariate Time Series Classification of Chemical Sensors
  • Preprint 2023
  • Semantic transfer of molecular representation chemistry corpus with early classification of multivariate time series
• Model-Agnostic Meta Learning for Remote Sensing
  • Deep Learning - Computer Vision
• Benchmarking Policy Gradient and DQN-Family Reinforcement Learning Models on Discrete and Continuous Games
  • Reinforcement Learning
• CASIS LLNL 2024 Vision-Language LLMs From Scratch Tutorial Seminar
• CASIS LLNL 2024 Computational Objectness from Motion in Video Poster and Oral
• Stanford ICME Transformers and Large Language Models Workshop

Education and Awards

• Worcester Polytechnic Institute, Worcester Massachusetts
  • Ph.D. - Data Science 2021 - 2023. 4.0/4.0 GPA
  • M.S. - Data Science 2019 - 2021. 4.0/4.0 GPA
  • $90,000 Annual Presidential Fellowship Award
• Reed College, Portland Oregon
  • Bachelor of Arts - Mathematics and Statistics 2016 - 2019
  • Presidential Commendation for Excellence
  • Dean’s List 2016 - 2017