Arnab Samanta

I am a data scientist and machine learning engineer with hands-on experience building production GenAI and NLP systems. I have developed end-to-end products that combine backend APIs, vector databases, and frontend interfaces to deliver measurable business impact.

At Citi, I built an enterprise GenAI product to automate vendor fraud model governance, reducing draft time by 65% and eliminating regulatory error scope. My implementation work spans Django backends, NoSQL databases, containerization with Docker, and deployment on GCP.

During my internship at Trademo, I automated HS Code classification workflows, designed a 12-language translator and vector database, and built recommender systems using Siamese networks, HNSW+PQ, and ranking-reranking approaches to reach strong top-3 accuracy.

I build practical ML products — from AI mock-interview agents and finance research tools using large LLMs and vector search, to delivery partner optimization pipelines — and prioritize robust engineering, reproducibility, and deployment-ready solutions.