Job Description

DRYFT

Ideas only change the world when you build them at scale. That’s what Dryft is here for.

Dryft builds models that automate complex human decisions in manufacturing operations. We capture the knowledge of experienced operators across supply chain, production planning, engineering and more, simulate the consequences of their decisions, and use this understanding to help run their business. Our customers include some of the largest robotics, machine and component manufacturers in the world.

We’re a small, talent-dense team with modest roots and a love for the physical world: cars, machines, planes. Our team includes graduates from Stanford, Berkeley, CDTM and TUM, winners of Stanford TreeHacks, recipients of Fulbright & Haniel scholarships, and people who have built everything from robots and race cars to software with millions of downloads.

What matters more to us than credentials is how people work: humility, hard work, speed, kindness, deep customer care, and a willingness to take on problems others avoid. Working at Dryft is hard. You get unusual ownership early on, with real responsibility. But that’s also what makes it one of the biggest learning experiences of your life.

We’re backed by founders and operators behind some of the most legendary tech companies of our time, Tier-1 U.S. venture firms, and board members of global industrial companies.

ERPs ran the physical world for half a century. Dryft is building the intelligence layer that will power the next.

​

THE ROLE

As a Full-Stack Engineering Intern, you'll build the system that decides how some of the largest manufacturers in the world operate. That means combining agent-based reasoning with mathematical optimization and simulation on large-scale industrial data.

• Build agents people trust – when an agent runs a factory's week, the largest industrial companies in the world have to trust it with their critical processes and understand every decision it makes; you build the routing, orchestration and guardrails behind that
• Ship features end-to-end – you understand the user need, and you get the freedom to build it across the LLM pipeline, infrastructure, backend and frontend, in production in your first weeks
• Work close to the users – sit with the planners who use it, see what would make their work 10x faster, and build it

You'll own your own features from day one, and what you ship decides what a plant does next week.

​

YOU

• Full-stack shipper – you've shipped production code end-to-end, you build scalable backend and infra that handles massive amounts of structured and unstructured data, and when you touch the frontend it feels like magic
• Highly technical – you've built something technically hard, and you can walk us through every decision behind it
• Product sense – you spot the feature hiding in a user's complaint, and you've deleted your own favorite one because nobody used it
• You care about getting it right – you know when it has to be 100% and when to keep it simple, and when it matters, you're still chasing the root cause at 2am
• Sense of urgency – you'd rather ship it tonight than sleep on an open PR
• Relocation – you're ready to work onsite in Munich or San Francisco (we support visa)

​

TECH STACK

• FE (TypeScript): React, Chakra UI v3, TanStack Router, TanStack Query, Vite
• BE (Python): FastAPI with Pydantic v2, Temporal, Postgres with pgvector, dagster
• AI: Pydantic AI, Langfuse