Ayush Gupta

Machine Learning Engineer with experience building end-to-end ML systems, including retrieval-augmented generation (RAG) pipelines, multiagent systems, and LLM-based applications. Skilled in data preprocessing, model optimization, and deploying scalable ML solutions using Python and modern ML frameworks.

Machine Learning Intern at AllSoft Solutions Ltd. (An IBM Partner) (2023-06 – 2023-07)

• Built an end-to-end churn prediction model for the E-commerce/DTH domain, addressing account-level churn impacting multi-customer revenue
• Performed data preprocessing, feature engineering, and EDA to identify key churn drivers (service score, tenure, customer care interactions, revenue trends)
• Handled imbalanced data using SMOTE-Tomek and trained Logistic Regression, Decision Tree, Random Forest, and Neural Network (MLP) models using scikit-learn
• Optimized models with GridSearchCV, achieving ~88% test accuracy with the Neural Network while maintaining strong precision, recall, F1-score, and ROC-AUC
• Proposed a revenue-safe, segmented retention strategy targeting high-risk accounts without blanket discounts, aligned with Revenue Assurance constraints
• Designed complete ML pipeline including preprocessing, training, and evaluation

R&D Intern at LPU Innovation Studio (2023-07 – 2023-08)

• Participated in innovation and intellectual property training, gaining exposure to patent filing processes and IP fundamentals
• Conducted research-driven projects focusing on Machine Learning applications in Emotion Detection, including literature review and comparative analysis of ML approaches
• Collaborated in a cross-functional team to design and implement a forge counter mechanism, contributing to improved production accuracy and operational efficiency
• Applied analytical thinking and prototyping methodologies to bridge theoretical research concepts with practical implementation

B.Tech in Computer Science and Engineering — ML/AI Specialization – Lovely Professional University (2025)