Niloufar Eghbali

Professional Experience

Graduate Research Assistant

Michigan State University, East Lansing, MI
Human Augmentation and Artificial Intelligence Lab (HAAIL, Github)
May 2021 - Present
Conducted independent research in the field of machine learning, focusing on reinforcement learning, computer vision, and natural language processing.
Selected projects:

• [NLP] Led research projects in collaboration with Henry Ford Hospital to solve various NLP problems such as text summarization, correlation analysis, and text classification to enhance radiology workflow. Projects mainly focused on text representation and fine-tuning Large Language Models (T5, Falcon).
• [RL] Developed reinforcement learning approaches to optimize treatment strategies in intensive care units.
  – Focused on personalizing sedation and delirium management, enhancing patient care through data-driven decision-making.
  – Combined RL with conformal prediction to make safer decisions for managing mechanical ventilation in ICU, resulted in 85% survival rate.
• [Computer Vision] Enhanced the Vision Transformers' performance using learnable radiomics features for improved feature extraction and image representation in medical image segmentation.

Graduate Teaching Assistant

Michigan State University, East Lansing, MI
Section Coordinator - Course: Algorithmic Thinking and Programming
Jan 2021 – May 2021

• Led group discussions and delivered presentations on "Algorithmic Thinking and Programming".
• Coordinated and supervised the activities of 8 fellow undergrad teaching assistants, providing guidance to ensure adherence to the curriculum and effective teaching methodologies.

Software Engineer

Dotin, Tehran, Iran
Core banking developer and designer
Feb 2015 – Aug 2020

• Collaborated in designing and developing critical core banking software modules, ensuring operational efficiency for Iran's major banks and financial service institutions.
• Played a role in a dynamic, cross-functional team to develop innovative solutions for customer needs.
• Participated in Iran’s largest banking software project: migrating over 20 million customer records to the new core banking system, ensuring a seamless transition for clients.
• Served as a code reviewer, mentoring junior developers to enhance code quality and functionality.

Graduate Research Assistant

Tarbiat Modares University, Tehran, Iran
Research assistant under the supervision of Dr. Gholam Ali Montazer
Aug 2015 – Aug 2016

• Conducted research, focusing on the enhancement of multiclass classification algorithms using Error Correcting Output Codes (ECOC).
• The research led to a model that was more accurate and required fewer classifiers than traditional random sparse ECOC approaches (published paper in the Pattern Recognition Letters, 2017).