Lead Engineer, LLMs
About Atomic Tessellator
We are building a sovereign future by empowering companies and nations to achieve materials breakthroughs, enhance product performance, and secure supply-chain independence through the development of the most powerful first-principles simulation platform for engineering materials research.
We're a seed-stage company with a headcount of five, and have been around for a little over a year. In this time, we have:
- Built a distributed worker architecture to modularise computational materials science operations.
- Scaled machine-learned interatomic potential (MLIP) models to enable multi-GPU inference, letting us model up to 700,000 atoms
- Completed pilot projects across aerospace, defence, nuclear fusion, and advanced polymers.
- Discovered (and are in the process of patenting) two materials, one of which is a high-temperature rare earth magnet substitute.
How we operate
- Speed is our bread and butter, bias-to-action is our competitive edge.
- We're constantly error-correcting, and balance explore/exploits.
About the role
Atomic Tessellator is a virtual lab that's capable of discovering new materials entirely within simulation. The foundations are built, now our goal is to make this system recursively exponential.
We're seeking a Lead LLM engineer to help accelerate our materials discovery pipelines and take our platform to the next level.
Your role would entail building a system for us to replicate experiments materials research papers within Atomic Tessellator, and automate the creation of materials research pipelines. This means that you'll be working at the absolute frontier of technology, between the intersection of language models and materials AI.
- You'll have complete ownership of the LLM-facing interfaces, e.g. CLI or MCP server - your decisions and your tastes.
- You'll have access to unlimited LLM credits from whichever providers of your choosing.
- You'll be working closely to support our scientists with hypothesis generation and validation.
- You'll also be developing internal tooling and coding development workflows for the entire team - meaning you'll need familiarity with concepts like loops and frequent intentional compaction.
Developments in the LLM space are frequent and chaotic - but what we've seen is a slow convergence towards ground truths of how to best effectively utilise these tools. Ideally you should be able to see past the trends, understand this, and build for where the puck is going.
We understand this is a very open-ended role, as such we don't expect much experience. We're looking for examples of projects where you've demonstrated the hacker trait.