Skip to content
View Priyanshu-byte-coder's full-sized avatar

Highlights

  • Pro

Block or report Priyanshu-byte-coder

Report abuse

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

Report abuse
Typing SVG
   

 About

name       : Priyanshu Doshi
location   : India
education  : BTech in AI & ML, Nirma University

domains:
  - Self-Supervised Learning & Fault Detection
  - Graph Neural Networks & Fraud Detection
  - Facial Emotion Recognition & Wellbeing AI
  - Content-Based Recommendation Systems
  - Explainable AI & Digital Twins
  - Full-Stack Web Development

currently:
  learning  : [ "Advanced PyTorch", "Transformers", "GenAI" ]
  building  : [ "SSL fault detection research", "Offline AI systems" ]

 Featured Projects

AI / ML / Research


Robotic Arm Fault Detection Self-supervised learning for robotic arm fault detection using the CASPER dataset. Benchmarked against SVM, Logistic Regression, Naive Bayes, CatBoost, and QDA with full statistical analysis across 10 runs. Python Self-Supervised Learning CatBoost scikit-learn

Fraudulent Review Detection Heterogeneous GNN using GraphSAGE and HeteroConv with Focal Loss, hard example mining, and burst detection. Includes web app for real-time review analysis on 121K+ test samples. PyTorch PyG GNN GraphSAGE

Maya — Wellbeing AI 100% offline AI companion on Raspberry Pi 5. Fuses facial emotion recognition, VADER sentiment, and Microsoft Phi-3 Mini via Ollama with ChromaDB RAG memory for empathetic conversations. Python Ollama FER ChromaDB Flask

Movie Recommender Content-based filtering using cosine similarity with real-time poster fetching via TMDb API. Deployed on Streamlit Cloud. Python Streamlit scikit-learn TMDb API

PyTorch Learning Structured notebooks covering PyTorch fundamentals — tensors, autograd, neural network building, and deep learning experiments. PyTorch Jupyter Deep Learning

Full-Stack / Hackathons


LUMIN.AI — Solar Risk Monitor Full-stack platform predicting solar inverter failures using XGBoost, Predictive Digital Twins, SHAP-based XAI, and Groq Llama 3.3 70B for plain-English explanations. AWS RDS + LangSmith observability. Next.js XGBoost SHAP LangChain FastAPI

MZHub — AI Spiritual Platform Marketing website for an AI-powered spiritual platform connecting temples, ashrams, churches, and mosques. Built with Next.js App Router, glassmorphism UI and dark theme. Next.js TypeScript Framer Motion

Secure Pharmacy System C++ prescription management preventing fake prescriptions with role-based access control (Patient/Doctor/Chemist), OTC classification, and audit trails. C++ OOP File Handling

Voice Bot AI-powered voice interaction application built with Google AI Studio for natural conversational experiences. Python Google AI Studio

GDG Farmer Solution Google Developer Groups Solution Challenge addressing agricultural challenges through technology. GDG Solution Challenge

Portfolio Personal portfolio with dark theme, creative animations, and responsive design. React TypeScript Vite

 Tech Stack


 AI / ML & Data Science

Python PyTorch scikit-learn Pandas NumPy Jupyter XGBoost MATLAB Ollama LangChain

 Languages & Frameworks

C C++ JavaScript TypeScript React Next.js Flask FastAPI Streamlit

 Tools & Platforms

Git GitHub Google Cloud AWS Postman VS Code Vercel Raspberry Pi

 GitHub Analytics


Top Languages



Streak Stats



Activity Graph

 Pinned Repositories


     

 Contribution Snake


Snake animation

 Connect


         


Built with curiosity — Priyanshu Doshi

Pinned Loading

  1. CatBoost_Anomaly_Detection_implementation CatBoost_Anomaly_Detection_implementation Public

    Python 1

  2. mzhub mzhub Public

    TypeScript 1

  3. pytorch_learing pytorch_learing Public

    Journey of learning pytroch

    Jupyter Notebook

  4. Mitanshp5/MECup Mitanshp5/MECup Public

    Python 1

  5. movie-recommendor-showcase movie-recommendor-showcase Public

    A content based movie recommender system using cosine similarity,get recommendation by the movies that you have already seen

    Jupyter Notebook

  6. Neal006/Fantastic4 Neal006/Fantastic4 Public

    TypeScript 2