Sumit Sahu

I’m a data science intern currently building AI and machine learning applications, focused on practical outcomes like smarter screening and more reliable loan monitoring. I enjoy turning messy data into models that can detect issues early and support better decisions.

In my current role, I built an AI agent to monitor loan offers before rollout, analyzing counts and detecting drift using LangGraph, LangChain, Python, and Streamlit. I also developed a machine learning model to identify drivers behind declining loan disbursements in a segment using ML techniques and data preprocessing.

Before that, I created an NLP-based application to automate subjective question checking using Word2Vec and achieved 86.2% accuracy. My projects include resume screening with TF-IDF and cosine similarity, transformer fine-tuning on customer data, and a RAG-based AI lawyer using FAISS and sentence transformers for semantic search.