Aryan V S

Hello stranger. Welcome to my corner of the internet.

Carl Sagan once said "We are a way for the cosmos to know itself." We are, indeed. My life's ambition is to build machines that can understand the cosmos and, in doing so, improve our own understanding.

Right now, my personal interests revolve around programming at the lowest levels. Eventually, I want to go deeper: understanding the electrical, mechanical, and materials engineering that make modern computers work at all. The entire stack, from silicon to intelligence. Maybe understanding the full silicon stack, alongside biological evolution, is how we eventually put artificial souls into rocks.

After all, we've already managed to fool rocks into thinking. I want to understand how that happened at the level where things almost don't work at all: how transistors got so tiny, how unreliable signals are coerced into stable logic, how timing, noise, power draw, and heat constrain design choices, and how layers of (fragile?) assumptions and abstractions were stacked on top of one another until computation emerged. Not the abstract diagrams from school textbooks, but the details where the different worlds of engineering collide.

I hold a simple belief: what I cannot build, I probably don't understand very well. This is why, if you know me IRL, you'll find me reversing things and writing my own implementations constantly. It's the best way to truly learn how something works. (Though, please don't do this in production code—your coworkers will hate you.)

I'm not much of a writer, but I wanted a place to collect thoughts and disperse them to interested readers. That place is here on the blogs page. Posts are organized by maturity: Seed for raw ideas and scattered thoughts, Sprout for concepts taking shape, Sapling for well-developed content, and Blossom for fully matured posts with comprehensive information.

News

Experience

M

Staff, Stealth Startup Sep 2025 - Present

Anything and everything to build the best Video World models.

Scout, Andreessen Horowitz Sep 2025 - Present

Investing in AI and Infra startups. Finding strong founders tackling hard problems.

Seeking pre-seed/seed angel checks? Reach out!

Staff, Hugging Face Jul 2024 - Sep 2025

Core maintainer of Diffusers and Finetrainers.

Research Intern, Microsoft Jan 2024 - Jul 2024

Computer vision and human-computer interaction research.

Real-time inference optimization for diffusion, GAN, NeRF, and Gaussian Splatting models in facial avatar and scene reconstruction settings. Privacy-focused user state understanding.

Advisor: Dr. Venkata Padmanabhan

Engineer, Modelslab Oct 2023 - Jan 2024

Inference optimization for generative models (image/video diffusion, voice cloning, LLMs).

Managed production GPU deployments, improved platform stability, and supported enterprise clients.

Application Development Intern, Morgan Stanley May 2023 - Jul 2023

Migrated internal financial tooling to SpringBoot and Angular. Optimized SQL queries.

Machine Learning Intern, Uptrain Mar 2023 - Apr 2023

Helped build Uptrain's (YC W23) open-source toolkit.

Talks

• IIT Indore - Aug 2025, "Careers in ML". Career guidance for undergrads.
• MLADS Conference - Jul 2024, "Hybrid Hallway". Presented and live-demoed Hybrid Hallway, and early explorations into real-time avatars and privacy-focused user state understanding, to Microsoft research leadership, professors, and hundreds of attendees.
• Microsoft Research India - Mar 2024, "Hybrid Hallway". Presented and live-demoed Hybrid Hallway.