A lot of my work has been around privacy, trust, and product systems at Viven, so I can share how I think about building features like access controls, evaluations, and workflows in real products. If that sounds useful, I’m happy to dig into it.
About
At Viven, Kusum builds privacy and trust features across the Digital Twin platform — including the Privacy Rule Management System with LLM-generated summaries, a Personas feature for team-level privacy controls, multi-recipient Scheduled Queries, and an evaluation framework for Digital Twin agents using Recall@N and MRR@N metrics. She joined as a Software Engineer Intern in summer 2025, where she designed a runtime Filter Query system for privacy enforcement and automated email-based access workflows in React, Flask, and TypeScript.
On campus she works as an Undergraduate Research Assistant on Protein Fold Classification (refining ML reconstruction modules) and earlier on a research project parsing TeX from 100,000+ papers to build a structured equation dataset. She has also been a Teaching Assistant for Object-Oriented Programming I in Java at UMD, supporting 150+ students. She graduates in May 2026.
