I'm not just a coder; I'm an Innovator dedicated to scaling bleeding-edge AI solutions from the research bench to enterprise-grade web applications. My expertise is a rare, powerful fusion of MERN Stack Mastery and Published Deep Learning Research.
| 🌟 Deep-Tech & AI Dynamo | 💻 Full-Stack & Cloud Architect |
|---|---|
| 6+ Research Papers in IEEE/International Journals (Strabismus Diagnosis, Disaster-Net, Steganography). | MERN Stack Master: Built and scaled complex applications for 500+ active users. |
| Specialized in Quantum ML (Penny Lane), Steganography, and Multimodal Transformers. | Architected RESTful APIs (Node.js/Express/Flask) and deployed with CI/CD pipelines. |
| Developed 'Machine Chatbot Pro' RAG system at Mahindra & Mahindra, boosting efficiency by 30%. | Proven efficiency: Cut deployment time by 60% and optimized systems for 100+ simultaneous users. |
- 🔭 Research Focus: I am currently pushing the boundaries in Robust Steganography and AI-Powered Medical Image Security/Classification.
- 📚 Deep Learning Stack: TensorFlow, Keras, PyTorch, OpenCV, and Penny Lane (Quantum ML).
- 💻 Tech Stack Focus: Deep-diving into Next.js, Three.js, and GSAP for high-performance, immersive web experiences.
- 🏆 Proven Drive: 200+ LeetCode problems conquered.
| Role | Company/Institution | Key Achievement |
|---|---|---|
| Summer Intern | Mahindra & Mahindra Ltd. | Developed RAG-based AI Chatbot (70%+ accuracy), reducing average issue resolution time by 30%. |
| Full-Stack Developer | Needle AI (Startup) | Architected and deployed scalable MERN applications, implementing CI/CD that reduced deployment time by 60%. |
| Research Assistant | Thapar University | Developed Deep Learning models for medical imaging; achieved 90%+ accuracy on the MIAS dataset; implemented Hybrid Classical-Quantum ML. |
| Publication Title | Focus | Key Result |
|---|---|---|
| Scalable Ensemble Framework for Robust Image Steganography (Accepted at INSPECT 2025) | Hybrid Traditional-Neural Steganography, Confidence Fusion. | 99.56% Accuracy for 8-256 bit payloads; PSNR up to 54.95 dB. |
| Dual-Stream Adaptive Fusion for Building Damage Segmentation (Journal Paper) | Satellite Remote Sensing, DeepLabV3, ResNet50. | Achieved 63.17% mIoU on the BRIGHT benchmark across 7 global disaster types. |
| MAW-Gen: AI-Powered Adaptive Watermarking Framework (Patent-Pending) | GAN-based attack resistance, Wavelet/Feature-space embedding. | Invisible, multi-bit watermarks for AI-content certification and medical protection. |
I wield the full power of the MERN stack alongside advanced AI and DevOps tooling to build robust systems.
Let's discuss how my expertise in scalable, robust AI and full-stack architecture can deliver your next breakthrough product.
- Email:
bhumit07205@gmail.com - Portfolio: https://bhumitg07205.github.io/New-Portfolio/