Skip to content
View I-Deepanshu's full-sized avatar

Block or report I-Deepanshu

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
I-Deepanshu/README.md
Deepanshu Shimar — AI & ML Engineer

Typing SVG



GitHub   LinkedIn   Gmail


Medium   LeetCode   ORCID


🤖 About Me

Hi! I'm a B.Tech CS (AI & ML) graduate passionate about building production-grade intelligent systems that solve real-world problems.

  • 🤖 ML Engineer at Rechk — BERT paper intelligence (82.4% acc, ROC-AUC 0.946)
  • 🛰️ Research Intern at NIT Warangal — Crop yield ML (LightGBM R²=0.906)
  • 🛡️ ML Intern at DRDO — Toxic gas classifier (XGBoost 95.02%)
  • 📄 Published Researcher — 2 ML papers
  • 🌟 Microsoft Learn Student Ambassador
  • 🏆 Top-60 Finalist — FHIR® Hackathon 2025
  • 🥇 Amazon ML Challenge 2025 — Top 1200 National Rank
  • 🎓 UIET, MDU · 2022–2026

🛠️ Tech Stack

Languages

ML / DL

NLP & LLMs

Tools & Cloud

📊 GitHub Stats

Streak Stats

Streak

GitHub Stats

Stats

Top Languages

Langs


📈 Contribution Graph

Activity Graph

🐍 Contribution Snake

Snake Dark

💻 Commit Activity

Profile Details


Repos per Language   Most Commit Language   Productive Time


🚀 Featured Projects

⚖️ IPC & BNS Predictor — RAG Legal AI

LangChain FAISS Gemini 3.5 Flash HuggingFace

RAG system mapping FIRs to 500+ IPC / 350+ BNS sections at 92% accuracy. Cut manual review by 60% and response time by 40%.

Demo

📄 Rechk — AI Research Paper Classifier

BERT FastAPI Docker Gemini 3.5 Flash

End-to-end pipeline: ingestion → BERT quality (Q1–Q4) → grammar correction → plagiarism detection. 82.4% accuracy on 1,350+ papers.

Live

🥔 Potato Disease Detection

Swin MobileViT Grad-CAM Streamlit FastAPI

Benchmarked 7 transformer models on 717 leaf images. 97% accuracy (Swin), 94.86% (MobileViT, 5.6M params). Grad-CAM interpretability.

Code

🛒 Amazon ML Challenge 2025 — Multimodal Price Prediction

PyTorch Qwen3 DINOv2 Cross-Attention

Cross-attention fusion of Qwen3 (text) + DINOv2 (vision). Top 1200 national rank, SMAPE 50–52% via 5-fold CV + ensemble stacking.

Code


📚 Research Publications

# Paper Venue
1 Solar Panel Degradation Prediction using ML — Stacked ANN, XGBoost & RF on 480-day UK PV data; best ensemble R² > 0.96 ResearchSquare Paper Code
2 Fuel Cell Degradation Prediction Using ML Models — Extra Trees Regressor (MAE=0.00099, R²=0.996) on PEM Fuel Cell Dataset ResearchSquare Paper Code

Dev Quote


💙 Thanks for visiting my profile!

Feel free to connect and collaborate.

🚀 Keep learning, Keep growing.


Pinned Loading

  1. INSIGHT_ONCO INSIGHT_ONCO Public

    TypeScript 2

  2. Fuel-cell-degradation-predictor Fuel-cell-degradation-predictor Public

    2

  3. IPC IPC Public

  4. Potato-Disease-Detection Potato-Disease-Detection Public

    Jupyter Notebook

  5. Solar-Panel-Degradation-Predictor Solar-Panel-Degradation-Predictor Public

    Jupyter Notebook 1