Skip to content
View Devashish-Rawat1's full-sized avatar
💭
Grind is on
💭
Grind is on
5 followers · 9 following

Block or report Devashish-Rawat1

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
Devashish-Rawat1/README.md

> Devashish Rawat

Data Scientist · ML Engineer · Builder of Things That Actually Work

Typing SVG

whoami

devashish = {
    "role"      : "Data Scientist / ML Engineer",
    "focus"     : ["Forecasting & Optimization", "Computer Vision", "Deep Learning"],
    "currently" : "Building production-ready ML pipelines & CV systems",
    "fun_fact"  : "I also shipped a Chrome Extension on Chrome Store",
}

⚡ Featured Projects

📦 DemandSense

ML-Powered Demand Forecasting & Inventory Optimization

End-to-end retail intelligence system — from raw sales data to actionable inventory decisions. Trained 4 models on 913K rows of retail history, deployed as an interactive dashboard.

  • Best model: XGBoost (RMSE 7.91 / MAPE 12.44%)
  • SHAP explainability for business-ready insights
  • EOQ, Safety Stock & Reorder Point planner
  • Batch prediction via CSV upload
  • Live on Streamlit Cloud

XGBoost LightGBM LSTM SHAP Streamlit Plotly Pandas

🎨 StyleFusion-HF

Real-Time Neural Style Transfer via AdaIN

Implemented the Huang & Belongie (2017) AdaIN architecture from scratch. VGG-19 encoder extracts features; a custom decoder reconstructs stylized images. Two-phase training curriculum on Kaggle.

  • Arbitrary content + style image pairs
  • Alpha slider to control stylization intensity
  • Real-time inference via Gradio
  • Deployed on Hugging Face Spaces

PyTorch VGG-19 AdaIN Gradio Hugging Face Computer Vision

⭐ Path-To-Anime

Chrome Extension — Anime Discovery Companion

A full-featured Chrome Extension (Manifest V3) for anime discovery. Smart filters by genre, rating, episode count; streaming platform links; personalized recommendations; and a personal watchlist.

  • Smart multi-filter search engine
  • Direct links to Netflix, Crunchyroll, HiDive
  • Published on Chrome Web Store
  • My Anime List with shareable watchlists

JavaScript HTML CSS Chrome Extension Manifest V3

// more in the works...

$ git status
On branch main
Changes staged for commit:
  - new_project_1.py  [NLP pipeline]
  - new_project_2.py  [Generative AI]
  - new_project_3.py  [MLOps tooling]

Stay tuned. 👀

🛠 Tech Stack

Languages

Python C++ JavaScript HTML5 CSS3 SQL

ML / Deep Learning

PyTorch TensorFlow Keras XGBoost LightGBM scikit-learn

Data & Visualization

Pandas NumPy Seaborn Matplotlib SHAP

Deployment & Tools

Streamlit Flask Hugging Face Jupyter Git GitHub Kaggle Anaconda

GitHub Stats

GitHub Streak

Connect

GitHub Instagram Hugging Face Chrome Store

"The model is only as good as the engineer behind it."

vibing

Pinned Loading

  1. DemandSense DemandSense Public

    ML-powered demand forecasting + inventory optimization. XGBoost, LightGBM & LSTM models · SHAP explainability · EOQ/Safety Stock planner · Streamlit dashboard.

    Jupyter Notebook 2

  2. Path-To-Anime-v1.2 Path-To-Anime-v1.2 Public

    Path-to-Anime – Your Ultimate Anime Discovery Companion Tired of endless scrolling? Whether you're new to anime or a long-time fan, Path-to-Anime helps you find your next binge-worthy show in secon…

    JavaScript 2

  3. StyleFusion-HF StyleFusion-HF Public

    Real-time arbitrary neural style transfer using AdaIN and VGG-19, served via a Hugging Face Spaces. Upload any content + style image pair and blend results with an alpha slider.

    Python 2