연구부터 프로덕션까지, Verena를 만드는 사람들. Verena는 의료 영상부터 대규모 언어 모델까지 모델을 직접 학습시키는 연구자들이 만들며, 첫 범위 설정부터 프로덕션까지 작업에 끝까지 책임을 집니다.
-
Highlights
- B.S. Computer Science & Mathematics @ University of Florida (2024-2027)
- Volunteer Researcher @ MIRTH AI Lab (2024-Present)
- OPS Employee @ Computational Microscopy Imaging Laboratory (2025-Present)
Research Interests
- Self-supervised learning.
- Diffusion models and flow matching.
- Reinforcement learning.
- Geometric constraints and feature representations (manifolds, spectral structure).
- Current focus: flow matching for medical image synthesis and geometric representation learning.
Timeline
Volunteer Researcher, MIRTH AI Lab
- Period: 2024 - Present
- Location: Gainesville, FL
- Trained diffusion and flow-matching generative models for high-fidelity medical image synthesis.
- Developed implicit neural representations (INRs) for resolution-independent volumetric analysis.
OPS Employee, Computational Microscopy Imaging Laboratory
- Period: 2025 - Present
- Location: Gainesville, FL
- Contributing to a large NIH-funded grant focused on multimodal AI predictions for nephrology.
- Built self supervised abstract feature extraction pipelines across medical imaging modalities including ultrasound.
- Trained semantic embeddings from clinical reports using encoder-decoder architectures.
B.S. Computer Science, B.S. Mathematics, University of Florida
- Period: Aug 2024 - May 2027
- GPA: 3.95 / 4.00
- Relevant coursework:
- Deep Learning in Medical Image Analysis
- Linear Algebra for Data Science
- Real Analysis with Advanced Calculus I and II
Publications
Geometry-Aware Implicit Neural Reconstruction of Oblique Micro-Ultrasound Scans
AortaGPT: An Interactive Vision-Language System for Aortic CT Analysis
- Venue: Under review (2026)
-
Highlights
- Incoming M.S. Computer Science @ Rice University (Fall 2026)
- B.S. in Computer Science @ University of Florida (Aug 2022 - May 2026)
- Undergraduate Researcher @ MIRTH AI Lab (Jan 2025-Present)
Research Interests
- Large Language Models (LLMs) and Vision-Language Models (VLMs).
- Theoretical foundations of AI, information theory, and Reinforcement Learning (RLVF).
- High-Performance Computing (HPC) and distributed training (Data/Expert Parallelism).
- Current focus: Reinforcement learning for medical VLMs, and optimizing distributed model training.
Timeline
Undergraduate Researcher, MIRTH AI Lab
- Period: Jan 2025 - Present
- Location: Gainesville, FL
- Developing deep learning model for 3D MRI segmentation to quantify brain volumes in preterm infants, directly impacting clinical evaluations of antibiotic exposure.
- Implementing Reinforcement Learning pipelines using Group Relative Policy Optimization (GRPO) for Medical 3D Vision-Language Models (VLM).
Lead ML Engineer, GatorLM
- Period: Jan 2026 - Present
- Location: Gainesville, FL
- Building a full-stack AI chatbot and managing the project architecture for a custom-trained DeepSeek-style LLM.
- Developed and trained a 9B parameter Mixture of Experts (MoE) model from scratch.
- Leveraged NVIDIA Blackwell B200 GPUs on the HiPerGator cluster, utilizing Transformer Engine for FP8 training to maximize throughput.
- Implemented a distributed training strategy combining Data Parallelism, Expert Parallelism, and Optimizer Sharding to handle the model’s scale efficiently.
Teaching Assistant, University of Florida
- Period: Jan 2024 - Jun 2024 & Spring 2026
- Location: Gainesville, FL
- Mentored students in C++ and Python, reinforcing core concepts of memory management and data structures.
- Developed exams and technical assessments to test students programming skills and conceptual understanding.
B.S. in Computer Science, University of Florida
- Period: Aug 2022 - May 2026
- GPA: 3.59 / 4.00
- Relevant Coursework:
- Information Theory
- Deep Learning in Medical Image Analysis
- Linear Algebra
- Additional Coursework: Signals and Systems
Key Projects & Research
Medical AI Agent (Full Stack AI)
- Built an agentic framework integrating LLM orchestration with MONAI 3D segmentation tools.
- Engineered a sandboxed Python execution environment to allow the agent to run analysis code autonomously, bridging software design with deep learning applications.