• 🎓 CSE Undergrad at GSFC University (2022–2026)
• 🧪 Deep Learning Engineer specializing in Satellite Imagery (SAR → Optical), GANs, and Super-Resolution
• 💻 Full Stack Developer (Next.js, TypeScript, Node.js, MongoDB, SQL)
• 🛰 Passionate about AI research, spaceborne imaging, and developing high-impact tech
• 🎉 Built Ananta’25 Web Platform serving 5,000+ users
• 🛠 Strong foundation in DSA (Graphs, Trees, DP, Sliding Window, BFS/DFS)
• 🤝 Always open to collaborations on AI/ML, web apps, and research engineering

🔗 https://github.com/Arnavshah22/Dual-Image-Super-Resolution-for-optical-Satellite-Imagery

• Developed a dual-stream EDSR-based architecture processing two complementary satellite images independently before fusion.
• Implemented feature fusion using channel-wise attention + concatenation, improving spectral & structural consistency.
• Built complete PyTorch ecosystem: residual blocks, pixel-shuffle upsampling, dataset loaders, augmentations, evaluation metrics.
• Achieved 22.7 dB PSNR, outperforming classical and single-image SR baselines.
• Visualized results with bicubic → baseline → dual-SR comparisons for research interpretability.
• Engineered extensible fusion mechanisms enabling future researchers to plug-and-play new feature strategies.

(Based on your research paper)

• Designed a two-stage disentangled learning framework separating SAR noise from content structure.
• Stage 1 learns noise-aware latent encoding, removing speckle influence from semantic representations.
• Stage 2 produces high-fidelity optical images using a Content–Semantic Colorization Network trained with perceptual & reconstruction losses.
• Utilized hybrid losses: cycle-consistency, perceptual, adversarial realism, improving color stability across terrains.
• Achieved superior PSNR, SSIM & visual coherence compared to Pix2Pix/vanilla GAN baselines.
• Built full SAR preprocessing pipeline (normalization, speckle modeling, alignment).

🔗 Website- https://www.anantagsfcu.in/

• Led development of the official fest platform, used by 5,000+ students & faculty.
• Built modular system: Event Engine, QR Pass System, Sponsor Panel, Admin Console, Digital Vault.
• Backend using Node + Express + SQL, deployed on Vercel + Cloudflare achieving 99.9% uptime.
• Designed dynamic dashboards for volunteers, coordinators & admins with real-time data sync.
• Collaborated across Media, Sponsorship, Events & Logistics teams ensuring smooth integration.
• Managed deployment pipeline, version control, UI/UX, testing, and content workflows.

🔗 https://github.com/Arnavshah22/furniture-rental-2.0

• Built using Next.js 14 App Router, TypeScript, MongoDB, featuring real-time search & category-based filtering.
• Implemented secure JWT Auth + bcrypt, ensuring protected user sessions.
• Designed Cart & Checkout system, multi-item billing, state persistence, DB optimization.
• Developed an Admin Dashboard for product management, orders & customer queries.
• Added modern animations with Framer Motion, responsive layouts & reusable components.
• Structured the app using modular APIs, optimized MongoDB queries and schema indexing.

• Improved facial recognition attendance accuracy from 70% → 95% using caching & multithreading.
• Built NLP-powered resume scanning system for automatic shortlisting.
• Integrated ML pipelines with HR dashboards and automated workflows.

• Built maze-solving engine using A* and Dijkstra, achieving 98% optimal path accuracy.
• Implemented OpenCV + PyWavelets pipeline for maze preprocessing.
• Developed real-time dashboard with Flask for visualization (<2 sec latency).

📧 Email: arnavshaw2015@gmail.com