Loading Portfolio...

AI Research • Full-Stack Developer • Engineering Student

Bridging Deep Learning Research with Real-World Applications

Get In Touch View Projects

About Me

I'm a passionate B.E. student at Thapar Institute of Engineering and Technology, specializing in Electronics and Computer Engineering. My journey combines cutting-edge AI/ML research with practical full-stack development, creating innovative solutions that bridge the gap between academic research and real-world applications.

0
Research Papers
0
Projects
2023
Graduation Year

Education

🎓

Bachelor of Engineering

Electronics and Computer Engineering

Thapar Institute Of Engineering and Technology, Patiala

2023-2027

Research & Publications

Published on IEEE Xplore

Enhancing Strabismus Diagnosis with Deep Learning from Detection to Classification

AIMLA 2025

Detection and Classification of Strabismus Using Trained Models Of ML. Full pipeline: Data Annotation, Data Preprocessing, Augmentation, Training, Evaluation

ML Models Tensorflow Keras CVAT.AI Roboflow
Accepted in INSPECT 2025

A Scalable Ensemble Framework for Robust Image Steganography and Data Restoration Under Distortion Attacks

Conference Paper

Innovated hybrid traditional-neural steganography model supporting multi-bit payloads (8–256 bits) with 99.56% accuracy and PSNR up to 54.95 dB.

CNNs Gen-AI Transformer-based
Submitted to CVIP 2025

DisasterNet: Joint Learning of Tweet and Image Features for Damage Severity Classification

IIT Ropar

Designed a multimodal pipeline that processes both tweet text and associated disaster images for classifying damage severity levels.

ML Models Tensorflow Transformer Multimodal
Submitted to Signal, Image and Video Processing Journal

Dual-Stream Adaptive Fusion for Post-Disaster Building Damage Segmentation

Benchmark Solution • Computer Vision • 2025

Advanced dual ResNet architecture with dynamic adaptive fusion for robust pixel-level assessment of building damage from multimodal satellite images. Empowers rapid, scalable AI analysis for real-world disaster response.

Dual ResNet-50 Adaptive Fusion (Concat/Subtract/Attention) DeepLabV3 Focal Loss PyTorch BRIGHT Dataset
Submitted to ACM Journal

Machine Learning for Ultrasound Report Generation: A 2013–2025 Comprehensive Survey of Techniques, Challenges, and Translational Potential in Low-Resource Settings

Review Paper

Comprehensive review of ultrasound report generation using Machine Learning, Deep Learning, and Generative AI approaches over the past decade.

CNNs RNNs Transformer-based LLMs BERT GPT
Submitted as Book Chapter (CRC Press, "Generative AI for Next Gen Defence")

Pixels Meet Words: Multimodal Transformer Models for Disaster Information Detection

Conference Paper

Revolutionary multimodal AI framework harnesses BERT-ViT fusion to decode crisis communication signals, achieving 85% accuracy with superior precision-recall dynamics—proving that intelligent cross-modal reasoning trumps raw performance metrics in life-critical disaster response systems.

CNNs Gen-AI ViT Multimodal
Under Writting

MAW-Gen: AI-Powered Adaptive Watermarking Framework

Journal Paper

Formulated MAW (Multidimensional Adaptive Watermarking) - imperceptible watermarks with adaptive embedding across domains

CNNs RNNs Transformer-based LLMs ViT DCT/DWT GPT

Featured Projects

TAAS Website

Full Stack

Developed an official website for the Thapar Amateur Astronomers Society to showcase activities, photo galleries, and about events. Includes an Admin Panel to manage content, events, and messages dynamically

React.js Node.js GSAP THREE.js MongoDB

E-Commerce Website

Full Stack

Built a complete E-Commerce platform with User Logins, product listings, cart and authentication. Includes an Admin Dashboard for managing products, orders

Node.js React.js CSS MongoDB Auth

Apple UI/UX Clone

Frontend

Designed a visually stunning and responsive clone of the official Apple website. Focused on delivering smooth UI/UX interactions with modern frontend practices

React.js TailwindCSS GSAP Three.js Express.js

MAW: Metamorphic Adaptive Watermarking Platform – Invisible Protection for Digital AI Content

Full Stack

Pioneered web platform utilizing MAW technology for secure, AI-generated image provenance and content authenticity

React.js Node.js GSAP THREE.js MongoDB Flask Gen-AI Watermarking

Skills & Technologies

Interactive network of my technical expertise

Programming Languages
ML/AI Technologies
Web Development
Tools & Databases

Experience

Summer Intern

Mahindra & Mahindra Ltd

On-Site

Jun 2025 - Aug 2025

  • Full-stack development using Python Flask and React for Machine Chatbot Pro
  • Integrated local LLMs (Ollama) with FAISS and Sentence Transformers for intelligent Q&A over technical documents
  • Implemented JWT-based user authentication and managed MongoDB cloud database operations
  • Designed and developed modern, responsive chat UI optimized for multi-turn conversations and history mapping
  • Delivered a 24/7 AI-powered maintenance assistant with 70%+ response accuracy and 2–5s answer time
  • Optimized secure deployment for both cloud and on-premises environments

Under-Graduate Research Assistant

Dr. Debabrata Ghosh & Dr. Neeru Jindal

Thapar University

Jan 2025 - Present

  • Image annotation using Roboflow and CVAT AI
  • Worked with medical imaging datasets
  • Deep Learning Model Architectures: CNNs (Efficient NetB7, ResNet, AlexNet, VGGNet, VIT), SVM, LLMs
  • Image classification, object detection, data augmentation techniques
  • Natural Language Processing: Data preprocessing, BERT and Transformer-based models
  • Currently contributing to more research papers in ML and AI

Leadership & Involvement

Technical Executive

Saturnalia'25, Thapar University, Patiala

Aug 2025 - Present

  • Contributed to Saturnalia, Thapar University's flagship techno-cultural fest with 10000+ nationwide participants
  • Engineered interactive engagement robot with autonomous navigation, speech recognition, and real-time attendee interaction capabilities
  • Implemented sensor fusion algorithms and conversational AI to enable seamless human-robot interaction throughout the 4-day event
  • Developed robust hardware-software integration using Jetson, computer vision, and wireless communication protocols
0 Students Reached

Head Of HKS Society

Thapar University, Patiala

Sept 2023 - Present

  • Organize large-scale devotional events attended by over 2000 students
  • Collected Donations, Handled PR and MEDIA Department
0 Students Reached

Get In Touch

Let's Connect!

I'm always open to discussing new opportunities, research collaborations, or interesting projects. Feel free to reach out!

✉️ bhumit07205@gmail.com
📱 +91-8884854056
💼 LinkedIn Profile
Open to Work