🎓 First-year CSE @ Delhi Technological University 🤖 Deep Learning enthusiast | Full-stack explorer | DSA beast in training 🎸 Fun fact: I play the guitar like I train models — with layers and soul.

• 🔭 Currently building and fine-tuning models in:

  • 🧠 LLMs | 🗣️ NLP | 🎯 Computer Vision
  • ⚙️ Retrieval-Augmented Generation (RAG) with Transformers
  • 📊 Custom Object Detectors | Facial & Human Action Recognition

• 🌐 Exploring the MERN Stack and leveling up my web dev game

• 📚 Deep diving into Data Structures & Algorithms with C++ for that extra edge

"I don’t just learn — I backprop through life." "I’m pretrained on hustle, fine-tuned on purpose."

"Every epoch makes me better"\

Languages:
💻 C++ | Python | JavaScript | HTML | CSS | C

Frameworks & Tools:

• 🧠 PyTorch, TensorFlow, OpenCV, Scikit-learn, HuggingFace
• 🌐 Node.js, Express, React, MongoDB
• 🧾 Git, GitHub, VS Code, Kaggle, Postman

• 🎵 Text-to-Music Generator with MusicGen & DiffSinger
• 🔍 RAG-based QA Bot with custom retrievers & LLMs
• 📦 Building object detectors: RetinaNet | EfficientDet | DETR
• 🧠 Facial Expression + Human Action Recognition (YOLO + CNN hybrid)
• 🌐 MERN-based Web Dashboards for ML Ops

• 🌐 GitHub
• 📬 Email: divansumishra47@gmail.com
• 🧠 Always open to collabs in AI, CV, or full-stack builds

       

• Every loss is just another gradient to descend
• I'm pretrained on red flags — I detect 'em in one shot
• YOLO? Nah. I detect chances in one shot
• Got 99 problems but vanishing gradients ain’t one
• Drop the past like a residual — I skip connections to the future
• I backpropagate through failures and optimize for growth
• Overfitting? Nah, I’m just deeply trained
• Even the best models carry some bias — so do we