Nodar Daneliya

Open a terminal, write a Rust web service, and ask for a database. With most clouds, that sentence ends in a week of YAML, dashboards, and a credit card you regret. With Shuttle, it ends in seconds. The database is just there, declared inside the code that needs it. That is the trick Nodar Daneliya has spent since 2019 perfecting: infrastructure that provisions itself from the application, instead of the application waiting on infrastructure.

Shuttle today runs underneath more than 20,000 developers and over 120,000 deployments. Its open-source core sits near the top of the Rust ecosystem on GitHub. And in October 2025 it raised a $6 million seed round to chase a bigger idea than Rust - the idea that AI can now write the code, but somebody, or something, still has to ship it.

That gap between "the AI wrote it" and "it is live and not on fire" is where Daneliya has planted his flag.

Most founders of a Rust developer-tools company come up through the obvious door: years of writing systems software, a chip on the shoulder about deployment, a side project that ate their weekends. Daneliya came in through a different one entirely.

He studied economics at UCL, then finance at Harvard, then a master's in financial analysis at London Business School. He passed through Google. He advised at The Raine Group. And then he went where the math was sharpest - leading trading at a machine-learning-driven quantitative hedge fund, building data pipelines and models that had to run in production, in real time, with real money on the line.

That is the unfair advantage hiding in his resume. A quant fund teaches you that the model is the easy part. The infrastructure - the pipes, the latency, the thing that has to never fall over - is the hard part. Daneliya carried that lesson out of finance and pointed it straight at the broader software industry, where millions of developers were fighting the same fight with worse tools.

In 2019 he co-founded Shuttle. In 2020 it went through Y Combinator's S20 batch. The bet underneath it has barely changed since: writing code was never the bottleneck. Running it was.

There is a seductive promise floating around software right now: describe an app in plain English, and an AI builds it. Tools like Lovable, Cursor and Replit have made the first half of that promise feel real. Code appears. It even looks right.

Then the developer hits the wall everyone forgets to mention. The code has to be deployed. It needs a database, secrets, a domain, a region, a bill. The AI that wrote a thousand lines in thirty seconds goes quiet, because shipping is a different sport than writing.

Daneliya's read is that this is not a bug in the AI wave - it is the opening. Shuttle assesses the deployment options, presents infrastructure as packages with pricing, and handles the cloud provider so the human, or the AI agent, never has to. The "spec" he describes is the translation layer: a thing humans can still review and AI can still reason about.

His plan is to push Shuttle past Rust into every language and every AI coding system. The company started narrow on purpose - Rust developers are demanding, and if it works for them, it works. Now the borders between ecosystems are dissolving, and he intends to be the layer that catches whatever the AI throws over the wall.