OpenCV • Medical Imaging • Image Segmentation • Object Detection • Image Classification • FLAIR MRI Processing • Dermatoscopic Analysis • OpenCV • Medical Imaging • Image Segmentation • Object Detection • Image Classification • FLAIR MRI Processing • Dermatoscopic Analysis • OpenCV • Medical Imaging • Image Segmentation • Object Detection • Image Classification • FLAIR MRI Processing • Dermatoscopic Analysis • OpenCV • Medical Imaging • Image Segmentation • Object Detection • Image Classification • FLAIR MRI Processing • Dermatoscopic Analysis • OpenCV • Medical Imaging • Image Segmentation • Object Detection • Image Classification • FLAIR MRI Processing • Dermatoscopic Analysis • OpenCV • Medical Imaging • Image Segmentation • Object Detection • Image Classification • FLAIR MRI Processing • Dermatoscopic Analysis • OpenCV • Medical Imaging • Image Segmentation • Object Detection • Image Classification • FLAIR MRI Processing • Dermatoscopic Analysis • OpenCV • Medical Imaging • Image Segmentation • Object Detection • Image Classification • FLAIR MRI Processing • Dermatoscopic Analysis • OpenCV • Medical Imaging • Image Segmentation • Object Detection • Image Classification • FLAIR MRI Processing • Dermatoscopic Analysis • OpenCV • Medical Imaging • Image Segmentation • Object Detection • Image Classification • FLAIR MRI Processing • Dermatoscopic Analysis • OpenCV • Medical Imaging • Image Segmentation • Object Detection • Image Classification • FLAIR MRI Processing • Dermatoscopic Analysis • OpenCV • Medical Imaging • Image Segmentation • Object Detection • Image Classification • FLAIR MRI Processing • Dermatoscopic Analysis •
TensorFlow • PyTorch • Keras • Transformers • CNN/ResNet • RNN/LSTM • Vision Transformers • UNet Architectures • FastAPI • Hugging Face • TensorFlow • PyTorch • Keras • Transformers • CNN/ResNet • RNN/LSTM • Vision Transformers • UNet Architectures • FastAPI • Hugging Face • TensorFlow • PyTorch • Keras • Transformers • CNN/ResNet • RNN/LSTM • Vision Transformers • UNet Architectures • FastAPI • Hugging Face • TensorFlow • PyTorch • Keras • Transformers • CNN/ResNet • RNN/LSTM • Vision Transformers • UNet Architectures • FastAPI • Hugging Face • TensorFlow • PyTorch • Keras • Transformers • CNN/ResNet • RNN/LSTM • Vision Transformers • UNet Architectures • FastAPI • Hugging Face • TensorFlow • PyTorch • Keras • Transformers • CNN/ResNet • RNN/LSTM • Vision Transformers • UNet Architectures • FastAPI • Hugging Face • TensorFlow • PyTorch • Keras • Transformers • CNN/ResNet • RNN/LSTM • Vision Transformers • UNet Architectures • FastAPI • Hugging Face • TensorFlow • PyTorch • Keras • Transformers • CNN/ResNet • RNN/LSTM • Vision Transformers • UNet Architectures • FastAPI • Hugging Face • TensorFlow • PyTorch • Keras • Transformers • CNN/ResNet • RNN/LSTM • Vision Transformers • UNet Architectures • FastAPI • Hugging Face • TensorFlow • PyTorch • Keras • Transformers • CNN/ResNet • RNN/LSTM • Vision Transformers • UNet Architectures • FastAPI • Hugging Face • TensorFlow • PyTorch • Keras • Transformers • CNN/ResNet • RNN/LSTM • Vision Transformers • UNet Architectures • FastAPI • Hugging Face • TensorFlow • PyTorch • Keras • Transformers • CNN/ResNet • RNN/LSTM • Vision Transformers • UNet Architectures • FastAPI • Hugging Face •
Deep Learning • Machine Learning • Neural Networks • Computer Vision • Natural Language Processing • Reinforcement Learning • Evolutionary Algorithms • Deep Learning • Machine Learning • Neural Networks • Computer Vision • Natural Language Processing • Reinforcement Learning • Evolutionary Algorithms • Deep Learning • Machine Learning • Neural Networks • Computer Vision • Natural Language Processing • Reinforcement Learning • Evolutionary Algorithms • Deep Learning • Machine Learning • Neural Networks • Computer Vision • Natural Language Processing • Reinforcement Learning • Evolutionary Algorithms • Deep Learning • Machine Learning • Neural Networks • Computer Vision • Natural Language Processing • Reinforcement Learning • Evolutionary Algorithms • Deep Learning • Machine Learning • Neural Networks • Computer Vision • Natural Language Processing • Reinforcement Learning • Evolutionary Algorithms • Deep Learning • Machine Learning • Neural Networks • Computer Vision • Natural Language Processing • Reinforcement Learning • Evolutionary Algorithms • Deep Learning • Machine Learning • Neural Networks • Computer Vision • Natural Language Processing • Reinforcement Learning • Evolutionary Algorithms • Deep Learning • Machine Learning • Neural Networks • Computer Vision • Natural Language Processing • Reinforcement Learning • Evolutionary Algorithms • Deep Learning • Machine Learning • Neural Networks • Computer Vision • Natural Language Processing • Reinforcement Learning • Evolutionary Algorithms • Deep Learning • Machine Learning • Neural Networks • Computer Vision • Natural Language Processing • Reinforcement Learning • Evolutionary Algorithms • Deep Learning • Machine Learning • Neural Networks • Computer Vision • Natural Language Processing • Reinforcement Learning • Evolutionary Algorithms •
Python • Pandas • NumPy • Scikit-learn • Matplotlib • Seaborn • Statistical Analysis • Python • Pandas • NumPy • Scikit-learn • Matplotlib • Seaborn • Statistical Analysis • Python • Pandas • NumPy • Scikit-learn • Matplotlib • Seaborn • Statistical Analysis • Python • Pandas • NumPy • Scikit-learn • Matplotlib • Seaborn • Statistical Analysis • Python • Pandas • NumPy • Scikit-learn • Matplotlib • Seaborn • Statistical Analysis • Python • Pandas • NumPy • Scikit-learn • Matplotlib • Seaborn • Statistical Analysis • Python • Pandas • NumPy • Scikit-learn • Matplotlib • Seaborn • Statistical Analysis • Python • Pandas • NumPy • Scikit-learn • Matplotlib • Seaborn • Statistical Analysis • Python • Pandas • NumPy • Scikit-learn • Matplotlib • Seaborn • Statistical Analysis • Python • Pandas • NumPy • Scikit-learn • Matplotlib • Seaborn • Statistical Analysis • Python • Pandas • NumPy • Scikit-learn • Matplotlib • Seaborn • Statistical Analysis • Python • Pandas • NumPy • Scikit-learn • Matplotlib • Seaborn • Statistical Analysis •
Academic Writing • Literature Review • Experimental Design • IEEE Publications • Research Methodology • Medical AI Research • Academic Writing • Literature Review • Experimental Design • IEEE Publications • Research Methodology • Medical AI Research • Academic Writing • Literature Review • Experimental Design • IEEE Publications • Research Methodology • Medical AI Research • Academic Writing • Literature Review • Experimental Design • IEEE Publications • Research Methodology • Medical AI Research • Academic Writing • Literature Review • Experimental Design • IEEE Publications • Research Methodology • Medical AI Research • Academic Writing • Literature Review • Experimental Design • IEEE Publications • Research Methodology • Medical AI Research • Academic Writing • Literature Review • Experimental Design • IEEE Publications • Research Methodology • Medical AI Research • Academic Writing • Literature Review • Experimental Design • IEEE Publications • Research Methodology • Medical AI Research • Academic Writing • Literature Review • Experimental Design • IEEE Publications • Research Methodology • Medical AI Research • Academic Writing • Literature Review • Experimental Design • IEEE Publications • Research Methodology • Medical AI Research • Academic Writing • Literature Review • Experimental Design • IEEE Publications • Research Methodology • Medical AI Research • Academic Writing • Literature Review • Experimental Design • IEEE Publications • Research Methodology • Medical AI Research •
Reinforcement Learning • Deep Q-Networks • Genetic Algorithms • Game AI Development • Neural Network Training • Reinforcement Learning • Deep Q-Networks • Genetic Algorithms • Game AI Development • Neural Network Training • Reinforcement Learning • Deep Q-Networks • Genetic Algorithms • Game AI Development • Neural Network Training • Reinforcement Learning • Deep Q-Networks • Genetic Algorithms • Game AI Development • Neural Network Training • Reinforcement Learning • Deep Q-Networks • Genetic Algorithms • Game AI Development • Neural Network Training • Reinforcement Learning • Deep Q-Networks • Genetic Algorithms • Game AI Development • Neural Network Training • Reinforcement Learning • Deep Q-Networks • Genetic Algorithms • Game AI Development • Neural Network Training • Reinforcement Learning • Deep Q-Networks • Genetic Algorithms • Game AI Development • Neural Network Training • Reinforcement Learning • Deep Q-Networks • Genetic Algorithms • Game AI Development • Neural Network Training • Reinforcement Learning • Deep Q-Networks • Genetic Algorithms • Game AI Development • Neural Network Training • Reinforcement Learning • Deep Q-Networks • Genetic Algorithms • Game AI Development • Neural Network Training • Reinforcement Learning • Deep Q-Networks • Genetic Algorithms • Game AI Development • Neural Network Training •
Docker • GitHub Actions • Docker • GitHub Actions • Docker • GitHub Actions • Docker • GitHub Actions • Docker • GitHub Actions • Docker • GitHub Actions • Docker • GitHub Actions • Docker • GitHub Actions • Docker • GitHub Actions • Docker • GitHub Actions • Docker • GitHub Actions • Docker • GitHub Actions •
Jupyter • VSCode • Google Colab • Jupyter • VSCode • Google Colab • Jupyter • VSCode • Google Colab • Jupyter • VSCode • Google Colab • Jupyter • VSCode • Google Colab • Jupyter • VSCode • Google Colab • Jupyter • VSCode • Google Colab • Jupyter • VSCode • Google Colab • Jupyter • VSCode • Google Colab • Jupyter • VSCode • Google Colab • Jupyter • VSCode • Google Colab • Jupyter • VSCode • Google Colab •
Team Leadership • Adaptability • Problem Solving • Communication • Collaboration • Team Leadership • Adaptability • Problem Solving • Communication • Collaboration • Team Leadership • Adaptability • Problem Solving • Communication • Collaboration • Team Leadership • Adaptability • Problem Solving • Communication • Collaboration • Team Leadership • Adaptability • Problem Solving • Communication • Collaboration • Team Leadership • Adaptability • Problem Solving • Communication • Collaboration • Team Leadership • Adaptability • Problem Solving • Communication • Collaboration • Team Leadership • Adaptability • Problem Solving • Communication • Collaboration • Team Leadership • Adaptability • Problem Solving • Communication • Collaboration • Team Leadership • Adaptability • Problem Solving • Communication • Collaboration • Team Leadership • Adaptability • Problem Solving • Communication • Collaboration • Team Leadership • Adaptability • Problem Solving • Communication • Collaboration •

Hi, I'm Shubham Gajjar

AI Researcher & M.S. Artificial Intelligence Student at Northeastern University, exploring deep learning, computer vision, and biomedical imaging.

Download ResumeGet In Touch

About Me

I'm an AI researcher and graduate student at Northeastern University, passionate about advancing deep learning applications in biomedical imaging and computer vision. My work focuses on hybrid neural architectures for tumor segmentation and skin lesion classification, bridging healthcare and AI.

I've published my research on "Hybrid ResNet-ViT for Skin Cancer Classification" at the 4th IEEE World Conference on Applied Intelligence and Computing (AIC 2025), achieving 96.3% accuracy and Area Under Curve of 1.00 across all classes. I've also contributed to research currently under review at Elsevier, focusing on VGG16-MCA UNet for brain tumor segmentation. My work demonstrates state-of-the-art performance in medical image analysis, and I'm currently exploring the integration of deep learning with healthcare and biological imaging as part of my ongoing research journey.

Published at IEEE AIC 2025
96.3% Model Accuracy Achieved
Download My CV

Technical Skills

Specialized expertise in AI/ML, computer vision, and research methodologies

AI/ML Core

Deep Learning
Machine Learning
Neural Networks
Computer Vision
Natural Language Processing
Reinforcement Learning
Evolutionary Algorithms

Deep Learning Frameworks

TensorFlow
PyTorch
Keras
Transformers
CNN/ResNet
RNN/LSTM
Vision Transformers
UNet Architectures
FastAPI
Hugging Face

Computer Vision

OpenCV
Medical Imaging
Image Segmentation
Object Detection
Image Classification
FLAIR MRI Processing
Dermatoscopic Analysis

Data Science & Analytics

Python
Pandas
NumPy
Scikit-learn
Matplotlib
Seaborn
Statistical Analysis

Research & Development

Academic Writing
Literature Review
Experimental Design
IEEE Publications
Research Methodology
Medical AI Research

Game AI & RL

Reinforcement Learning
Deep Q-Networks
Genetic Algorithms
Game AI Development
Neural Network Training

Cloud & DevOps

Docker
GitHub Actions

Tools

Jupyter
VSCode
Google Colab

Leadership & Adaptability

Team Leadership
Adaptability
Problem Solving
Communication
Collaboration

Research Publications

Published research in medical AI and computer vision, contributing to healthcare advancement

Hybrid ResNet-ViT for Skin Cancer Classification

Shubham Gajjar, Harshal Joshi, Om Rathod, Vishal Barot, Deep Joshi

Accepted

4th IEEE World Conference on Applied Intelligence and Computing (AIC 2025)

2025 • DOI: 10.1109/AIC60235.2025.11212073

Designed hybrid architecture combining frozen ResNet50 feature extractor with four-head Vision Transformer blocks, attaining 96.3% accuracy and macro F1 of 0.961 on HAM10000 dataset. Integrated Global Average Pooling and multi-head self-attention for seven-class skin lesion classification, achieving Area Under Curve of 1.00 across all classes. Accepted and presented at IEEE World Conference on Applied Intelligence and Computing (AIC 2025) to 100+ attendees.

Research Areas

Computer VisionDeep LearningMedical AISkin CancerResNetVision Transformer
View CertificateView on IEEE

VGG16-MCA UNet for Brain Tumor Segmentation

Shubham Gajjar, Deep Joshi, Avi Poptani, Vishal Barot

Under Review

Elsevier

2025 • DOI: Pending

Led innovation by designing VGG16-based encoder with Multi-Channel Attention decoder achieving 99.59% accuracy and 99.71% specificity on LGG Brain MRI Segmentation dataset from 110 low-grade glioma patients. Implemented ensemble learning combining multiple model configurations, improving Dice coefficient by 3.7% over standard UNet. Applied data engineering with preprocessing pipeline implementing skull stripping, intensity normalization, and resizing to 256x256 pixels for FLAIR MRI scans.

Research Areas

Computer VisionDeep LearningMedical AIBrain TumorUNetFLAIR MRIMulti-Channel Attention

Extended ResNet50 with Inverse Soft Mask Attention for Skin Cancer

Shubham Gajjar, Harshal Joshi, Om Rathod, Vishal Barot, Deep Joshi

Submitted

Journal Submission

2025 • DOI: Pending

Developed two-stage pipeline combining U-Net++ hair segmentation with Extended ResNet50 classifier featuring Inverse Soft Mask Attention mechanism, achieving 97.89% accuracy on HAM10000 dataset with 10,015 dermoscopic images. Applied creativity by integrating dense residual blocks and Squeeze-and-Excitation modules with learnable weighted feature aggregation for hair-occluded and unoccluded regions. Utilized Nadam optimizer with Cosine Decay Restarts and Sparse Categorical Crossentropy loss, incorporating explainable AI principles ensuring model deployment readiness.

Research Areas

Computer VisionDeep LearningMedical AISkin CancerResNetInverse Soft Mask AttentionU-Net++

Projects

Cutting-edge research in medical AI and innovative AI/ML projects showcasing deep learning expertise

Research
Accepted

Skin Cancer Classification (Research)

Medical AI Research - Accepted at IEEE AIC 2025

Designed hybrid architecture combining frozen ResNet50 feature extractor with four-head Vision Transformer blocks, attaining 96.3% accuracy and macro F1 of 0.961 on HAM10000 dataset. Integrated Global Average Pooling and multi-head self-attention for seven-class skin lesion classification, achieving Area Under Curve of 1.00 across all classes. Accepted and presented at IEEE World Conference on Applied Intelligence and Computing (AIC 2025) to 100+ attendees.

PythonTensorFlowPyTorchDeep LearningComputer VisionMedical AIResNet50Vision TransformerHAM10000 DatasetImage Classification
Research
Under Review

Brain Tumor Segmentation (Research)

Medical AI Research - Under Review at Elsevier

Led innovation by designing VGG16-based encoder with Multi-Channel Attention decoder achieving 99.59% accuracy and 99.71% specificity on LGG Brain MRI Segmentation dataset from 110 low-grade glioma patients. Implemented ensemble learning combining multiple model configurations, improving Dice coefficient by 3.7% over standard UNet. Applied data engineering with preprocessing pipeline implementing skull stripping, intensity normalization, and resizing to 256x256 pixels for FLAIR MRI scans.

PythonTensorFlowPyTorchDeep LearningComputer VisionMedical ImagingUNetVGG16Multi-Channel AttentionFLAIR MRI Processing
Research
Submitted

Extended ResNet50 with Inverse Soft Mask Attention for Skin Cancer (Research)

Medical AI Research - Submitted to Journal

Developed two-stage pipeline combining U-Net++ hair segmentation with Extended ResNet50 classifier featuring Inverse Soft Mask Attention mechanism, achieving 97.89% accuracy on HAM10000 dataset with 10,015 dermoscopic images. Applied creativity by integrating dense residual blocks and Squeeze-and-Excitation modules with learnable weighted feature aggregation for hair-occluded and unoccluded regions. Utilized Nadam optimizer with Cosine Decay Restarts and Sparse Categorical Crossentropy loss, incorporating explainable AI principles ensuring model deployment readiness.

PythonTensorFlowPyTorchDeep LearningComputer VisionMedical AIResNet50Inverse Soft Mask AttentionU-Net++HAM10000 DatasetImage Classification
AI/ML Core
Completed

TrackMania Reinforcement Learning Agent

Game AI & RL Research

Developed an advanced reinforcement learning agent for TrackMania racing game using Implicit Quantile Networks (IQN). The agent learns optimal racing strategies through trial and error, achieving competitive lap times and demonstrating robust decision-making in unpredictable racing situations.

PythonPyTorchReinforcement LearningImplicit Quantile NetworksNeural NetworksComputer Visiondxcam
View Code
Data Science
Completed

Twitter Sentiment Analysis (NLP Project)

Social Media Analytics

Built a comprehensive sentiment analysis system using Twitter API to analyze public sentiment on various topics. Implements NLP techniques and machine learning models for real-time sentiment classification.

PythonTwitter APINLPMachine LearningPandasNLTKTextBlobGoogle Colab
View Code

Certificates

Academic and professional certificates. Click to view or download.

AIC 2025 Presentation Certificate

IEEE AIC 2025 • 2025

View

Machine Learning Certificate

Northeastern University • 2024

View

Python Data Structures (Coursera)

Coursera

View

Python For Data Science

(Course)

View

Badges

Digital badges and credentials from Northeastern University

Essentials for Academic Success badge

Essentials for Academic Success

Northeastern University

View Credential
Foundations of Business Learning badge

Foundations of Business Learning

Northeastern University

View Credential
Foundations of Responsible AI Learning badge

Foundations of Responsible AI Learning

Northeastern University

View Credential
Foundations of Software Engineering and Data Management Learning badge

Foundations of Software Engineering and Data Management Learning

Northeastern University

View Credential

Let's Collaborate

Open to research collaborations in AI for healthcare, biomedical imaging, and computer vision.

Based at The Roux Institute, Northeastern University — Portland, Maine.

Research Focus

Medical AI, Computer Vision, Deep Learning

Expertise

Medical AI, Hybrid Deep Learning Architectures, Multi-Agent Systems, IEEE Publications

Connect With Me

LinkedIn

Professional network and research updates

GitHub

View AI/ML projects and research code

ORCID

Research profile and publications

Google Scholar

Academic publications and citations

Send a Message

Interested in collaborating on cutting-edge AI research or innovative machine learning projects? Let's connect!