Prashant Gupta

Data Scientist | Python, SQL, ML, Deep Learning, MLOps | ZS | IIT Roorkee

Gurugram, Haryana, India

About

The transition from academia to industry has been invigorating as I apply my Mechanical Engineering background from IIT Roorkee to drive data-led innovation at Bharat Petroleum. As a Data Analyst , I’ve leveraged analytical rigor and technical depth to enhance inventory management and customer satisfaction — achieving a 15% reduction in stockouts and a 30% improvement in reporting efficiency. My journey into advanced analytics deepened through developing CNN-based Next Best Action (NBA) models for rare disease brands (Gattex and Eohilia) under Takeda, where I executed 150+ model iterations that boosted campaign-led sales by +30% for Eohilia and +20% for Gattex. By automating data preprocessing, feature engineering, and validation pipelines in Python, I reduced model execution time by 80%, driving scalable and reliable solutions. Across every role, my goal remains consistent — to bridge business strategy and data science, collaborating cross-functionally to turn complex datasets into actionable insights that deliver measurable impact.

Experience

  • Decision Analytics Associate at ZS
    Jun 2025 - Present · 1 yr 2 mos

  • Data Analyst at Bharat Petroleum Corporation Limited
    Jun 2024 - May 2025 · 1 yr

    Analyzed fuel consumption data and implemented the SARIMA model, resulting in accurate forecasts that streamlined inventory management processes and reduced stockout occurrences by 15%. Automated reporting processes, reducing the time to generate key reports by 30%, enabling timely data-driven decision-making for senior leadership. Played a key role in cross-functional collaboration, contributing to strategic initiatives that improved customer satisfaction scores by 20%, while reducing operational delays by 12%.

  • Analytics Intern at Blume Global
    May 2023 - Jul 2023 · 3 mos

    Optimized Chatbot performance by analyzing user interactions to improve query resolution rates by 10%, reduce response times by 25%, which lead to overall user satisfaction. Processed large datasets for intent classification, boosting NLP model accuracy by 15%. Conducted A/B testing and predictive modeling to optimize chatbot workflows anticipating user needs.