Software Engineer, Agent Platform

We’re looking for a Software Engineer to build the agent platform that powers materials discovery, from the AI agents that design and run experiments to the applications scientists and customers use to work with them.

You’ll report to the Director of Software Engineering and work closely with the rest of the software engineering team. Your time splits between two kinds of work. The first is our agentic system: the agents that run campaigns and propose experiments, plus the evals and infrastructure that make them reliable in production. The second is shipping features end to end: giving an agent a new capability, evolving the campaign data model in Cortex (our Go backend), and building the screens in Tetra (our customer app) or Sentinel (our lab control panel) where people can review and steer the result.

This is not a research role, but you’ll work side by side with materials scientists, roboticists and hardware engineers to turn prototypes into production systems that run in our lab., * Build the agents that design materials discovery campaigns, proposing novel compositions that are made and measured in our lab

• Build the tools and workflows behind them, such as scientific literature search, simulations and data analysis in a sandbox
• Build the evals that measure our agents, from benchmarks to LLM judges, and hill-climb against them to improve agent quality
• Instrument agents with tracing, token and cost accounting, and structured logs so you can debug them in production
• Work across model providers and adapt our agents as new models change what they can do, * Evolve the Cortex data model for campaigns and experiments: protobuf schemas, MongoDB documents and migrations
• Build the pipelines that ingest and process scientific documents for our agents
• Ship each feature across every layer it touches: the data model, the agent, the API and the UI, * Write automated tests and evals on both sides of the API and contribute to CI/CD pipelines, * Experience with agent frameworks, durable execution and LLM observability tools, such as LangChain, Restate, Temporal or Langfuse
• Passion for agentic coding tools and keeping up with how fast the field changes
• Background in materials science, chemistry or physics
• Early-stage startup experience

What we offer

• Work closely with a team on the cutting edge of AI research
• A mission: an opportunity to fundamentally change the way humanity makes progress through materials science discovery

Core agent and backend expertise

• At least 5 years of professional experience building production backend systems
• Hands-on experience building and operating LLM-based agents or workflows in production: tool calling, structured outputs, context management, streaming and handling model failures
• Understanding of agentic design patterns, such as designer-critic loops and multi-agent systems
• Experience measuring agent quality with evals, test sets or LLM judges, and using the results to decide what to change

Frontend, backend and data fundamentals

• Professional experience with TypeScript and a modern frontend framework (Svelte/SvelteKit preferred; React or Vue acceptable if eager to learn Svelte), enough to ship a product feature on your own
• Experience designing and evolving API contracts (gRPC and Protocol Buffers preferred; REST or GraphQL acceptable)
• Working knowledge of databases, data modeling and running production data migrations (MongoDB or document stores preferred)
• Comfortable in a polyglot monorepo: our agents are in Python, Cortex is in Go, and our applications are in TypeScript

Communication and collaboration

• Clear, precise communicator, especially when working across disciplines
• Comfortable operating without a hard boundary between agent, backend and UI: you pick up whichever layer the feature needs, and you know when to pull in a specialist
• We actively use AI-assisted development and expect engineers to find their own productive workflows with these tools

The base salary range for this role is $165,000-$295,000. Where an offer lands depends on experience, skills, and scope.