Prateek Kumar

A Retrieval-Augmented Generation application that lets users ask natural language questions about their own documents and get context-aware answers. Handles document processing and chunking with LangChain, generates embeddings with Sentence Transformers, retrieves relevant context via a FAISS vector store, and uses an LLM to generate grounded responses — all through an interactive Streamlit interface for upload and querying.

An end-to-end pipeline that converts a website URL into a functional React Native (Expo) mobile app. Extracts UI structure via Playwright and generates native components through template-based code generation. Includes an LLM-based evaluator that scores generated app quality, with a human-in-the-loop review layer backed by SQLite to catch and correct generation errors before deployment.

A vulnerability-prioritization tool that computes an Exploitable Likelihood Score from CVSS and threat data, backed by a custom ServiceNow table to manage the underlying application data. Built UI Builder pages to visualize ELS scores, vulnerability groupings, and Security Incident records for security analysts.