Hi, I'm Manthan Hirani

A passionate researcher and developer focused on AI, Computer Vision, and Deep Learning. Explore my work or chat with my AI assistant to learn more.

View Projects

Development

Building robust applications with modern stacks like Next.js and Django.

Research

Deep diving into Deep Learning and Computer Vision focused on synthetic image analysis.

AI Integration

Implementing RAG-powered systems and LLM solutions for interactive experiences.

My Publications

Research Publications

Dynamic Risk Stratification in ICU Sepsis: A Deep Reinforcement Learning Approach Using Synthetic Data

Research Publication (IEEE REACS 2025): This study frames dynamic ICU sepsis risk stratification as a reinforcement learning problem using a Dueling Double Deep Q-Network (D3QN). Utilizing a privacy-safe synthetic dataset of 2,164 patient stays, the model evaluates 42 sequential clinical features and SOFA scores to predict patient improvement across 4-hour intervals. The trained agent attains 75% accuracy on unseen trajectories, capturing complex temporal patterns and demonstrating a clinically desirable bias toward correctly flagging non-improving cases for potential bedside decision support.

Reinforcement LearningHealthcare AIPython
My Projects

Featured Projects

Exploring the boundaries of AI, Computer Vision, and Deep Learning through research-led development.

Compact AI Image Detection

Developed an ultra-lightweight 55K parameter CNN for CIFAKE detection, achieving 95%+ accuracy. Performance parity with models 100x larger.

Computer VisionPyTorchResearch

Agentic AI Dashboard

Built an end-to-end GenAI dashboard with agentic capabilities and automated ETL pipelines for real-time business performance and data analytics.

LangChainStreamlitGenAI

AI Powered Chat Viewer

Created an AI-powered Chat Viewer web application using Streamlit to interact with and analyze large chat datasets efficiently.

NLPStreamlitPython

RAG Chatbot

Built a RAG-powered interactive portfolio for showcasing research papers using FAISS and LangChain for intelligent document retrieval.

RAGFAISSLangChain
My Recognition

Certifications & Achievements

Advance Course of Emerging Technologies

SAP and edunet

Learned emerging technologies like Machine Learning, IoT, Deep Learning, Computer Vision and SAP ABAP on BTP. Practical knowledge and implementation of various projects during 100 hours.

Fundamentals of Deep Learning

NVIDIA

Completed industry-recognized training covering core deep learning concepts including neural networks, CNNs, optimization techniques, and applications in computer vision and AI.

UGC-NET Qualified

Computer Science and Applications

Qualified National Eligibility Test (NET) for Assistant Professorship, demonstrating deep technical knowledge in CSE.

2024

GATE Qualified (DA)

Data Science and AI

Successfully qualified the Graduate Aptitude Test in Engineering in the Data Science and Artificial Intelligence paper.

2024

GATE Qualified (CS)

Computer Science and IT

Successfully qualified the Graduate Aptitude Test in Engineering in the Computer Science and Information Technology paper.

2024

About Me

Manthan Hirani

I am a dedicated researcher and developer with a deep focus on Computer Vision and Deep Learning. My work often explores the intersection of security and AI, particularly in synthetic image analysis.

With a strong foundation in both theoretical research and practical implementation, I bridge the gap between complex algorithms and user-centric applications. Whether I'm fine-tuning a neural network or crafting a pixel-perfect frontend, I strive for excellence and innovation.

Core Expertise

PyTorch
Hugging Face
LangChain
LLMs/SLMs
Computer Vision
Reinforcement Learning
FastAPI
FAISS
MCP
Docker
SQL/NoSQL
Natural Language Processing