ML/AI · SPATIAL COMPUTING · PRINCETON

I'm Ranveer

Building spatial intelligence

Studying world models and agentic AI @ Princeton

About

Hi! I'm Ranveer.

I'm a Junior at Princeton majoring in Computer Science with a focus on spatial intelligence: machine learning systems that understand and manipulate the physical world in 3D.

My current research focuses on relighting 4D Gaussian Splatting scenes, formal verification for hardware compilers, and improving LLMs for low-resource languages. Reach out if you're working on any of these or want to chat about building safe AI.

My other hobbies are cooking (worked in a Thai restaurant for 3 years), playing Magic: The Gathering, or making tech related instagram reels: @rsingh.daily.

Based inPrinceton, NJ

StudyingCS @ Princeton University

CurrentlyAI Engineering Intern @ Deloitte

Open toML research and fullstack engineering internships

Timeline

Experience

Research, infrastructure, and product engineering across Princeton and industry.

June 2026 — Present

AI Engineering Intern

Deloitte · Chicago, IL

Built and deployed agentic RAG services on Google Cloud Platform serving an internal application used by 20,000 employees, using Langfuse for latency and failure-mode observability.
Won 1st place in Innovation Challenge by creating AI powered safe compliance platform and presenting to audience of 300+.

PythonAzureLLMsLangChainFastAPIPyTorch

Selected work

Projects

Applied ML, research tools, and products built end to end.

swipe to explore

Product

Otto

A life assistant that lives in iMessage. Text Otto in to block distractions, log workouts to Apple Health, send emails and more.

Turns scattered thoughts into action over iMessage: focus shields, fitness sync, and calendar updates without a new dashboard to check.

AWSNode.jsPythonSwiftSentry

2026

Research

4D Spatiotemporal Relighting

First ever end-to-end relighting in 4D scenes (Gaussian Scenes with a moving object). Explores how light, materials, and geometry evolve across time to enable realistic illumination of moving objects.

Advances neural rendering for immersive AR/VR and digital twin pipelines where lighting must stay consistent as scenes change.

PyTorchNeural Rendering3D VisionCUDAPython

2025

Web Development

ProteinArchitect

A web platform for protein architecture analysis and visualization. Explore protein structures, analyze molecular interactions, and visualize complex biological data through an intuitive interface.

Built for researchers needing accessible structural biology tools without expensive software.

ReactTypeScriptPythonWebGLBioinformatics

2025

AI/Healthcare

DentalConnect

An AI-powered diagnostic platform for dental care that makes preventative care and accurate diagnostics more accessible. Patients can upload photos to receive instant visual reports highlighting potential issues like cavities, gingivitis, or discoloration.

Trained on 10,000+ dental images; built at a hackathon to make preventative care more accessible.

PythonReactFastAPIAWSSageMaker

2024

Writing

Blog

Notes on ML systems, product thinking, and building useful intelligence.

July 9, 2026 · 5 min

MIRA: Dreaming of Rocket League

A playable multiplayer world model of Rocket League. There's no game engine, just a neural net dreaming the next frame twenty times a second.

World ModelsAIResearch

Read article

July 6, 2026 · 8 min

What Exactly Are World Models

A brief history and explanation of world models, starting from dreaming agents to JEPA, Dreamer, and Genie. Exploring how imagination may lead us to AGI.

World ModelsAGI

Read article

March 14, 2026 · 8 min