Headline

AI Engineer | GenAI, LLMs, RAG, Agentic Systems | PyTorch | NITK

About

Data Scientist at NPCI with 1.5+ years of experience building AI-driven fraud detection systems for large-scale UPI transaction networks. I design and deploy scalable ML and deep learning models that detect financial fraud with high precision while minimizing false positives. My work includes building FP/FN analytics dashboards to improve model interpretability and decision-making, and implementing federated learning frameworks for secure, privacy-preserving collaboration with partner banks. Recently, I built an AI Expense Manager—an LLM-powered assistant that converts voice and natural language into structured financial records in real time (~90% accuracy, <500 ms latency), demonstrating my work in agentic AI systems and real-time NLP pipelines. Core strengths: • Fraud detection & risk modeling • LLMs, RAG pipelines & agentic AI systems • NLP (BERT & transformer-based models) • Scalable backend systems (FastAPI, SQL) • Production-grade ML systems & deployment

Location

Mumbai

Work experience

Data Scientist at National Payments Corporation Of India (NPCI), Sep 2024 - Present · 1 yr 8 mos, Mumbai, Maharashtra, India

Vice President at Dakshana Alumni Network(DAAN), Dec 2023 - Nov 2024 · 1 yr, Remote

Internship Coordinator at Career Development Centre (CDC) NITK Surathkal, Aug 2022 - Apr 2023 · 9 mos, On-site

Education

Bachelor of Technology - BTech, Computer Science, National Institute of Technology Karnataka

12th, Dakshana foundation

Public links

github: https://github.com/harshghandwani-ai