Rohit Gupta

Applied Scientist @ Amazon

About

Currently a graduate student at the Center for Research in Computer Vision at the University of Central Florida. My research is centered on multi-modal LLMs, with a special focus on improving video understanding capabilities of language models. Previously, I have also worked extensively on self-supervised and weakly-supervised image representations, adversarial robustness, image segmentation problems, video memorability and object re-identification.

Country

City

United States

Industry

Computer Software

Skill

Machine Learning, Data Mining, Python, R, Computer Vision, Web Development, Teradata, Matlab, Embedded Systems, C, C++, Distributed Systems, Robotics, Algorithms, Electrical Engineering, Quantitative Finance, Risk Modeling, Deep Learning

Experience

Applied Scientist

Amazon

2025-8 - Present · 1 yr 2 mos

Seattle, Washington, United States

Prime Video & Studios CoreTech

Graduate Research Assistant

UCF Institute of Artificial Intelligence (IAI)

2020-8 - 2025-7 · 5 yrs

Orlando, Florida, United States

• Current research has focused on Multi-modal LLMs, bridging Video and Text Domains for omni-capable models. • Developed robust representations for visual classification using self-supervised and supervised techniques. • Collaborated on research projects supported by DARPA, IARPA and NSF grants.

Applied Scientist Intern

Amazon

2024-5 - 2024-11 · 7 mos

Seattle, Washington, United States

Worked as part of Rufus-Multi-Modal LLM team for Stores Foundational AI group at Amazon Science. Internship project was primarily focused on building an unified Video LLM capable of unifying text and embedding generation to solve variety of video understanding tasks.

Applied Scientist Intern

Amazon

2023-5 - 2023-11 · 7 mos

Palo Alto, California, United States

Worked for M5 Team in the Search Science and AI group. Worked on open vocabulary video classification, training a model capable of identifying any actions and objects in videos, even if not seen during training. Submitted a paper based on work done to a top conference in the field.

Research Intern

SRI

2022-5 - 2022-8 · 4 mos

Menlo Park, California, United States

Developed multi-label, multi-modal prototype contrastive learning to achieve state-of-the-art results on the newly introduced task of fine-grained educational video classification. Work supported by NSF EAGER Award # 2139219: Technology to Review Online Videos for Education (TROVE)

Research Engineer

Conduent

2017-9 - 2019-7 · 1 yr 11 mos

Bengaluru Area, India

Worked on a variety of projects in Computer Vision: 1. Media memorability prediction 2. Analyzing multi-modal data for smart-city applications 3. Instance recognition and image classification for augmented reality (AR) applications 4. Appearance based re-identification of cars for traffic flow analysis

Data Scientist

Fuzzy Logix

2014-7 - 2015-6 · 1 yr

Bangalore

Responsible for contributing to the development of a suite of machine learning, statistical and financial algorithms (using C and SQL) embedded directly into massive parallel processing data warehouses like Teradata, Netezza, etc Created an open source R Package to serve as additional interface for the analytics suite Created a Python library for testing implementation of financial pricing models

Summer Analyst

Goldman Sachs

2013-5 - 2013-6 · 2 mos

Bangalore

Market Risk Modeling Team. Intern in a team responsible for making quantitative models to produce market risk metrics for a wide variety of instruments at the firm-wide level. Formulated and implemented a new model to efficiently estimate the market risk (stress tests and Value at Risk) caused due to FX volatility skew for firmwide portfolios of Goldman Sachs, containing million of securites