// available for opportunities
Aspiring ML Engineer and Python Developer with end-to-end ML pipeline experience. Passionate about building intelligent systems — from deepfake detection to predictive analytics.
01 — Who I Am
I'm Harshal Sonawane, an MCA student at KK Wagh Institute of Engineering & Research, Nashik, with a strong foundation in Machine Learning, Python development, and data-driven problem solving. I've completed two AICTE-recognized internships and built production-ready projects spanning deepfake detection, agricultural AI, and sports analytics. I'm currently gaining hands-on experience building full-stack LMS and CRM applications with Frappe/ERPNext. I love turning raw data into meaningful insights and deploying ML models that solve real-world problems.
02 — Toolkit
03 — Journey
Engineered full-stack LMS & CRM applications using Python and Frappe/ERPNext, implementing custom modules, REST APIs, and workflows following MVC architecture. Optimized backend logic across 3+ modules, improving data consistency and reducing manual workflow steps.
Built a Crop & Fertilizer Recommendation System using Random Forest (~88% accuracy on a 2,200-row dataset) and deployed a real-time Streamlit dashboard with a full EDA pipeline.
Developed 3 production-ready projects including a Resume, Themed Website, and Portfolio using HTML5 & CSS3, responsive across 5+ screen sizes.
Designed and developed a personal portfolio website using HTML, CSS, and JavaScript in collaboration with AICTE.
04 — Built Things
An end-to-end deepfake detection system built as MCA Final Year Project. Achieved 94.08% accuracy on benchmark deepfake datasets using CNN with transfer learning. Built with Python, OpenCV, Scikit-learn and deployed via Streamlit for real-time image/video authenticity prediction with confidence scores.
A live cricket score prediction system trained on 10,000+ match records achieving ~91% R² score. Powered by XGBoost with venue, team, over & wicket features engineered from historical data. Deployed as an interactive Streamlit dashboard with Pandas, NumPy, Matplotlib and Seaborn for visual analytics.
An AI-powered crop and fertilizer recommendation tool for farmers. Multi-class Random Forest classifier achieving ~88% accuracy on a 2,200-record dataset using 7 soil & climate features. Built with Scikit-learn, Pandas, Seaborn and deployed via Streamlit with full EDA and feature importance analysis.
05 — Verified
Machine Learning & AI Analyst — Symbiosis Institute (Jun 2024)
Supervised-unsupervised Learning, Model Evaluation, Feature Engineering
Python Programming — MKCL KLIC Course (2023)
Core Python, OOP, File I/O, Decorators, Generators, List Comprehensions
C++ Programming — MKCL KLIC Course (2023)
Object-Oriented Programming, Data Structures, File Handling
07 — Full Details
Resume preview goes here
08 — Let's Talk
