Pranjal Ballewar

I’m a final-year B.Tech (AI/ML) student who builds production-ready machine learning models, AI automation pipelines, and full-stack intelligent systems. I focus on turning data into measurable performance through strong preprocessing, evaluation, and deployable architecture.

In my Machine Learning Intern role at CODSOFT, I built Python-based classification and regression models using Scikit-learn and Pandas for telecom-adjacent financial datasets. I engineered end-to-end preprocessing pipelines (feature engineering, SMOTE, and hyperparameter tuning), reducing model error by ~18% and improving production reliability across 3 model iterations.

Through my projects, I’ve applied AI to real workflows and industry-relevant problems. I architected a Multi-Agent AI Automation System using CrewAI and LLaMA (Groq) with REST API ingestion and Streamlit delivery, and I built a Credit Card Fraud Detection system achieving >95% recall with low-latency production inference.

I’m also passionate about computer vision for automation, as shown in my Deep Learning Edge Detection pipeline. Using TensorFlow and Keras with OpenCV, I delivered a 5-stage training workflow and achieved 99.17% accuracy on 12 semantic edge types—backed by transfer learning, pseudo-labelling, and quantitative evaluation.