📍 Thane, Maharashtra | 🎓 B.E. Mechanical Engineering | 💼 3+ Years ML Experience
Machine Learning Engineer at WonderBiz Technologies (Schneider Electric) with expertise in Computer Vision, NLP, Reinforcement Learning, and AutoML. Passionate about building production-grade ML systems and open-source contributions. Co-inventor of an international AI patent for water management systems.
Active interests: Senior ML Engineer roles, Computer Vision Engineer, Time Series forecasting, and AI/ML YouTube content creation.
| Project | Description | Tech Stack | Status |
|---|---|---|---|
| Sudoku Intelligence Engine | Brain-inspired Sudoku solver using Maskable PPO. 100% solve rate on 1M+ puzzles with 1458-dimensional action space and custom Gymnasium environment. | Stable-Baselines3 Gymnasium PyTorch NumPy |
⭐ Star |
| NES Mario AI Trainer | Vision-based RL agent that learns to play NES Mario. Combines CNN-based feature extraction with policy learning. | TensorFlow OpenCV Gym |
⭐ Star |
| Pong RL Agent | Deep Q-Network agent learning to play Pong from raw pixel input. Demonstrates an end-to-end RL pipeline. | TensorFlow OpenAI Gym |
⭐ Star |
| Project | Description | Tech Stack | Status |
|---|---|---|---|
| Semantic Segmentation (MobileNetV2 U-Net) | Custom U-Net architecture for precise image segmentation. Leverages MobileNetV2 backbone for efficient inference. Includes Streamlit web interface. | PyTorch MobileNetV2 Streamlit |
⭐ Star |
| Auto Segmentation | Real-time semantic segmentation web app using MobileNetV2 + U-Net. Includes both CLI and Streamlit deployment. | PyTorch Streamlit PIL |
⭐ Star |
| Cancer Cell Segmentation (U-Net) | Custom U-Net for medical image analysis. Demonstrates biomedical image segmentation best practices. | PyTorch Jupyter |
🔬 Research |
| Vehicle Detection (YOLOv5) | Real-time object detection using YOLOv5 pretrained models with live screen capture. | YOLOv5 OpenCV |
⭐ Star |
| Animal Classification (CNN from Scratch) | Custom CNN trained on Microsoft dataset. Achieves >80% accuracy with optimised training. | TensorFlow OpenCV |
🎓 Learning |
| Optical Character Recognition (OCR) | Neural network-based OCR system with character recognition pipeline. | TensorFlow OpenCV |
⭐ Star |
| Project | Description | Tech Stack | Status |
|---|---|---|---|
| Orima — Time Series Analyzer | Interactive desktop app for time-series analysis. Auto-detects trends/seasonality, visualises ACF/PACF, recommends AR/MA/ARIMA/SARIMA models with plain-English explanations. | Python Tkinter Matplotlib Statsmodels |
🚀 Production |
| ARIMA/SARIMA Models | Empirical exploration of time-series forecasting techniques with VAR models. | Pandas Statsmodels Jupyter |
🔬 Research |
| Project | Description | Tech Stack | Status |
|---|---|---|---|
| Car Price Prediction | End-to-end Linear Regression pipeline with Flask web app. Full EDA → Feature Engineering → Model Training → Deployment. | Flask Scikit-learn Pandas MySQL |
⭐ Star |
| P2P Loans Risk Analysis | Credit risk modeling and loan classification using supervised learning. | Pandas Scikit-learn Scipy |
📊 Analytics |
| Book Recommendation System | Collaborative filtering-based recommendation engine. Demonstrates matrix factorisation concepts. | NumPy Scipy Pandas |
⭐ Star |
| Project | Description | Tech Stack | Status |
|---|---|---|---|
| Font Image Generator App | Desktop app for generating diverse character datasets from multiple fonts. Download v0.1 | Tkinter PIL |
🚀 Production |
| Code Mark Highlighter | VSCode extension for highlighting code blocks with persistent, content-aware tracking. | TypeScript VSCode API |
⭐ Star |
| One-Stop-Shop E-Commerce | Full-stack e-commerce platform with user authentication and product management. | Flask Python |
💼 Portfolio |
| Coin Tab Assignment | Data processing and analysis assignment. | Pandas NumPy |
📊 Analytics |
| Project | Description | Tech Stack | Status |
|---|---|---|---|
| Air Pollution Analysis: COVID-19 Impact | Interactive Tableau visualisation analysing pollution trends in India before/after COVID lockdown. | Tableau Data Analytics |
📊 Public |
| GitHub Profile Stats | Automated GitHub activity visualization and contribution analytics. | GitHub API Vercel |
📊 Public |
| Machine Learning | CNNs • U-Net • Transfer Learning • XGBoost • AutoML • Time Series (ARIMA/SARIMA) • Recommender Systems |
| Deep Learning | TensorFlow • PyTorch • Stable-Baselines3 • Computer Vision • NLP • Reinforcement Learning |
| Data Engineering | SQL • FastAPI • REST APIs • Docker • ETL Pipelines • RAG Systems |
| Data Science | Pandas • NumPy • Scikit-learn • Exploratory Data Analysis • Feature Engineering • Statistical Analysis |
| Deployment & Tools | Streamlit • Flask • Tableau • Git • FastAPI • Vercel |
| Languages | Python • SQL • JavaScript • TypeScript • HTML/CSS |
✅ International AI Patent Co-Inventor — "AI Based Water Management System" (App 19/214,503)
- Filed in: US, Australia, UAE, Chile, Oman, Peru
✅ Schneider Electric Global Hackathon — 🥈 2nd Place (50+ teams)
✅ ML Project Highlights:
- U-Net P&ID Segmentation — 93% accuracy on piping diagram segmentation
- AutoML Manufacturing Platform — Automated model selection and hyperparameter tuning
- Predictive Maintenance System — XGBoost-based vibration analysis (~10% reduction in unplanned downtime)
- RAG Question-Answering System — Built with Cohere API
I'm actively exploring Senior ML Engineer roles focusing on:
- 🤖 Applied ML Systems — Production ML pipelines and real-world problem solving
- 🔬 Computer Vision — Image analysis, segmentation, and object detection
- 📈 Time Series & Forecasting — Predictive analytics and anomaly detection
Always happy to discuss ML, collaborate on open-source, or explore exciting opportunities!
⭐ If you found something useful, don't forget to star it!
Made with ❤️ by Rumit Pathare