Profile banner
Suraj Prasad

Suraj Prasad

AI Researcher and Engineer focused on LLMs and computer vision

Mumbai, Maharashtra, India
Joined June 2025
Favicon imagein.linkedin.com
Favicon imagegithub.com
Favicon imagejarus77.github.io

Network

1.5K connections
🎓
IIT Students & Early Career Tech Professionals
📈
IIT Bombay Entrepreneurs & Business Leaders
🤖
AI/ML Engineers & Researchers
🚀
Startup Founders & EF Investors
🔧
General Engineering Students
💻
Aspiring Devs
💡
Growth Leaders

Summary

Suraj Prasad is a highly engaged student and researcher at IIT Bombay, specializing in Artificial Intelligence and Machine Learning. His academic pursuits involve foundational models, transformer architectures, and vision-language tasks, particularly within federated learning environments. github+1
He has a strong focus on practical applications of AI, demonstrated by his work on KuralGPT with BharatGen for Indic Languages and his contributions to the "MedPromptExtract" tool at Koita Centre for Digital Health, which addresses medical data extraction and anonymization. github+2
Suraj combines his technical skills with leadership and organizational roles, having served as a Manager and Computer Vision Engineer at Unmesh Mashruwala Innovation Cell and a Business Executive at the IITB Rocket Team (Aerove). His team achieved notable success in international competitions like Robodrive 2024 and Stranger Sections 2. github
His research contributions include co-authoring a paper on "Federated Cross-Modal Style-Aware Prompt Generation," which explores enhancing generalization in federated learning for vision-language models by integrating multi-scale visual features and style indicators. arxiv+2

Work

Education

Projects

Writing

Federated Cross-Modal Style-Aware Prompt Generation

August 1, 2025

A research paper introducing FedCSAP, a framework that harnesses multi-scale visual cues and domain-specific style variations to produce robust, context-aware prompt tokens for vision-language models in federated learning settings.

Favicon imagearxiv.org

MedPromptExtract (Medical Data Extraction Tool): Anonymization and High-Fidelity Automated Data Extraction Using Natural Language Processing and Prompt Engineering

July 1, 2025

A publication presenting MedPromptExtract, an automated tool utilizing NLP and prompt engineering to extract data from medical records like discharge summaries, while ensuring data anonymization.

Favicon imagepubmed.ncbi.nlm.nih.gov