Shubham Mantri

About Shubham Mantri

Academia IIT Roorkee (2017) – B.Tech, with a strong foundation in Data Science, ML, and Systems Design Experience 8+ years of hands-on experience in Data Science, Machine Learning, Deep Learning, and AI-driven applications Specialized in Computer Vision, Natural Language Processing, LLMs, RAG, and Agentic AI Currently working as Lead Data Scientist at Jio Platforms Ltd. Companies Experience Jio Platforms Ltd. – Leading innovation in Agentic RAG, multimodal systems, and Generative AI Bureau – Focused on Fraud Detection using ML and Behavioral Biometrics Spinny – Built hybrid ML systems for dynamic car pricing and procurement Skills Programming: Python, SQL, NoSQL Tools: Tableau, Excel, AWS, Redshift, Athena, MongoDB Systems: API Development (Flask, FastAPI), Airflow DAGs, System Design AI/ML Focus: NLP, CNNs, GANs, LLMs, Generative AI, Prompt Engineering, Agentic AI Libraries & Frameworks Data & Visualization: pandas, matplotlib, seaborn, streamlit ML/DL: scikit-learn, tensorflow, keras, pytorch NLP & LLM: spacy, transformers, langchain, llamaindex, langgraph Web & APIs: Flask, FastAPI Image/NLP Ops: rembg, openai, gradio Algorithms Expertise Supervised ML: Decision Tree, Linear Regression, Random Forest Deep Learning: CNN, GANs, MLP Anomaly Detection: Isolation Forest, ECOD AI Architectures: Naive RAG, Agentic RAG, Multimodal RAG LLM & Generative AI: Prompt Engineering, Retrieval-Augmented Generation Project Summaries Jio PDF Query Chatbot (Agentic RAG): Built an intelligent chatbot leveraging Agentic Retrieval-Augmented Generation to extract accurate insights from PDFs. Image-Style Attribute Mapping: Developed a system to match user-input attributes with visual elements, achieving 88% matching accuracy. Bureau Device Intelligence for Fraud Detection: Created device fingerprinting models using ML & rule-based logic. Achieved 99.4% accuracy, onboarding 10+ clients, and driving a 27% revenue uplift. Behavior Biometrics for Bot/ATO Detection: Used motion, touch, and pointer data with advanced anomaly detection models to reach 97% bot detection accuracy and 92% ATO fraud recall. Spinny Hybrid ML Pricing Engine: Designed a Decision Tree + Weighted Regression-based pricing model, achieving a 15% reduction in RMSE and optimizing procurement accuracy. Awards & Achievements \"Data Science Guru\"(Tech Sultan Award – Bureau) for pioneering work in Device Intelligence \"Best Analyst in Town\"(Spinny) for developing pricing models for second-hand cars

Experience

1. Lead Ai Engineer

   Nyse

   Feb 2026 — Present · Hyderabad, IN

Education

• Rukmani Birla Modern High School

  Class XII

  2012 — 2013

• Indian Institute of Technology, Roorkee

  Bachelor's degree

  2013 — 2017

• Rukmani Birla Modern High School

  Class X

  2010 — 2011