AI Engineer (interested in physical infrastructure)

Apply on YC · Original job listing

• Role: Engineering / Full stack
• Experience: 1+ years
• Visa: US citizen/visa only
• Skills: Python, Rust, TypeScript, Torch/PyTorch, Machine Learning, CAD Design, LLMs, RAG, AI Agents

Gradient is everything.

Torus is building the agentic engineering tool that physical engineering firms use to build infrastructure 10x faster. We turn design engineering work that took 6 weeks into 6 hours, and we're already working with multiple Fortune 500s and major EPCs on critical infrastructure like data centers and energy installations.

Who we are

Torus was founded in May 2026 by Marcus Lima (CEO) and Rahul Thayil (CTO). Marcus previously founded Heimdal (direct air carbon capture technology and infrastructure); Oxford and Cambridge grad. Rahul previously founded Orion (AI threat intelligence and response automation). We're a team of 6 in San Francisco, with domain-specialist engineers (electrical PEs, process engineers) who make the agent think like the right engineer for each seat. We started Torus because AI can finally do the structured, geometry-heavy work at the core of reindustrialization, and the industry is burning for it.

What you'll do

You'll do it all. This is a generalist seat: you own features end to end, from ingesting messy engineering inputs (CAD, drawings, spec packages) to shipping agent-generated outputs engineers actually trust. Our north star for where your attention goes: the agent, the data substrate it reasons over, the workspaces engineers work in, and getting real outputs into partners' hands. If you're not moving one of those four, ask yourself why.

We move fast. In your first week, your code is in production. By your first month, you'll ship something used daily (e.g. an extraction pathway going from prototype to production), own an end-to-end slice for one engineering discipline, and get your work into a Fortune 500 partner's workflow.